chore: modernize typing for Python 3.10+ minimum

py/src/braintrust/bt_json.py

@@ -2,7 +2,8 @@
 import json
 import math
 import warnings
-from typing import Any, Callable, Mapping, NamedTuple, cast, overload
+from collections.abc import Callable, Mapping
+from typing import Any, NamedTuple, cast, overload
 
 
 # Try to import orjson for better performance

py/src/braintrust/contrib/temporal/test_temporal.py

@@ -4,7 +4,7 @@
 import uuid
 from dataclasses import dataclass
 from datetime import timedelta
-from typing import Any, Dict
+from typing import Any
 
 import pytest
 import pytest_asyncio
@@ -31,7 +31,7 @@ class TestHeaderSerialization:
     def test_span_context_to_headers_with_valid_context(self):
         interceptor = BraintrustInterceptor()
         span_context = {"trace_id": "test-trace-id", "span_id": "test-span-id"}
-        headers: Dict[str, temporalio.api.common.v1.Payload] = {}
+        headers: dict[str, temporalio.api.common.v1.Payload] = {}
 
         result_headers = interceptor._span_context_to_headers(span_context, headers)
 
@@ -40,8 +40,8 @@ def test_span_context_to_headers_with_valid_context(self):
 
     def test_span_context_to_headers_with_empty_context(self):
         interceptor = BraintrustInterceptor()
-        span_context: Dict[str, Any] = {}
-        headers: Dict[str, temporalio.api.common.v1.Payload] = {}
+        span_context: dict[str, Any] = {}
+        headers: dict[str, temporalio.api.common.v1.Payload] = {}
 
         result_headers = interceptor._span_context_to_headers(span_context, headers)
 
@@ -78,7 +78,7 @@ def test_span_context_from_headers_with_valid_header(self):
 
     def test_span_context_from_headers_with_missing_header(self):
         interceptor = BraintrustInterceptor()
-        headers: Dict[str, temporalio.api.common.v1.Payload] = {}
+        headers: dict[str, temporalio.api.common.v1.Payload] = {}
 
         result = interceptor._span_context_from_headers(headers)
 

py/src/braintrust/devserver/schemas.py

@@ -1,8 +1,6 @@
 import json
 from collections.abc import Sequence
-from typing import Any, Union, get_args, get_origin, get_type_hints
-
-from typing_extensions import TypedDict
+from typing import Any, TypedDict, Union, get_args, get_origin, get_type_hints
 
 
 # This is not beautiful code, but it saves us from introducing Pydantic as a dependency, and it is fairly

py/src/braintrust/framework.py

@@ -17,13 +17,13 @@
     Any,
     Generic,
     Literal,
+    Protocol,
+    TypedDict,
     TypeVar,
-    Union,
 )
 
 from tqdm.asyncio import tqdm as async_tqdm
 from tqdm.auto import tqdm as std_tqdm
-from typing_extensions import Protocol, TypedDict
 
 from .generated_types import FunctionFormat, FunctionOutputType, ObjectReference
 from .git_fields import GitMetadataSettings, RepoInfo
@@ -216,13 +216,8 @@ class EvalScorerArgs(SerializableDataClass, Generic[Input, Output, Expected]):
     metadata: Metadata | None = None
 
 
-OneOrMoreScores = Union[float, int, bool, None, Score, list[Score]]
-OneOrMoreClassifications = Union[
-    None,
-    Classification,
-    Mapping[str, Any],
-    list[Classification | Mapping[str, Any]],
-]
+OneOrMoreScores = float | int | bool | None | Score | list[Score]
+OneOrMoreClassifications = None | Classification | Mapping[str, Any] | list[Classification | Mapping[str, Any]]
 
 
 # Synchronous scorer interface - implements callable
@@ -247,20 +242,19 @@ class AsyncScorerLike(Protocol, Generic[Input, Output, Expected]):
     async def eval_async(self, output: Output, expected: Expected | None = None, **kwargs: Any) -> OneOrMoreScores: ...
 
 
-# Union type for any kind of scorer (for typing)
-ScorerLike = Union[SyncScorerLike[Input, Output, Expected], AsyncScorerLike[Input, Output, Expected]]
+ScorerLike = SyncScorerLike[Input, Output, Expected] | AsyncScorerLike[Input, Output, Expected]
 
-EvalScorer = Union[
-    ScorerLike[Input, Output, Expected],
-    type[ScorerLike[Input, Output, Expected]],
-    Callable[[Input, Output, Expected], OneOrMoreScores],
-    Callable[[Input, Output, Expected], Awaitable[OneOrMoreScores]],
-]
+EvalScorer = (
+    ScorerLike[Input, Output, Expected]
+    | type[ScorerLike[Input, Output, Expected]]
+    | Callable[[Input, Output, Expected], OneOrMoreScores]
+    | Callable[[Input, Output, Expected], Awaitable[OneOrMoreScores]]
+)
 
-EvalClassifier = Union[
-    Callable[[Input, Output, Expected], OneOrMoreClassifications],
-    Callable[[Input, Output, Expected], Awaitable[OneOrMoreClassifications]],
-]
+EvalClassifier = (
+    Callable[[Input, Output, Expected], OneOrMoreClassifications]
+    | Callable[[Input, Output, Expected], Awaitable[OneOrMoreClassifications]]
+)
 
 
 @dataclasses.dataclass
@@ -278,27 +272,23 @@ class BaseExperiment:
     """
 
 
-_AnyEvalCase = Union[
-    EvalCase[Input, Expected],
-    EvalCaseDict[Input, Expected],
-    EvalCaseDictNoOutput[Input],
-    ExperimentDatasetEvent,
-]
+_AnyEvalCase = (
+    EvalCase[Input, Expected] | EvalCaseDict[Input, Expected] | EvalCaseDictNoOutput[Input] | ExperimentDatasetEvent
+)
 
-_EvalDataObject = Union[
-    Iterable[_AnyEvalCase[Input, Expected]],
-    Iterator[_AnyEvalCase[Input, Expected]],
-    Awaitable[Iterator[_AnyEvalCase[Input, Expected]]],
-    Callable[[], Union[Iterator[_AnyEvalCase[Input, Expected]], Awaitable[Iterator[_AnyEvalCase[Input, Expected]]]]],
-    BaseExperiment,
-]
+_EvalDataObject = (
+    Iterable[_AnyEvalCase[Input, Expected]]
+    | Iterator[_AnyEvalCase[Input, Expected]]
+    | Awaitable[Iterator[_AnyEvalCase[Input, Expected]]]
+    | Callable[[], Iterator[_AnyEvalCase[Input, Expected]] | Awaitable[Iterator[_AnyEvalCase[Input, Expected]]]]
+    | BaseExperiment
+)
 
-EvalData = Union[_EvalDataObject[Input, Expected], type[_EvalDataObject[Input, Expected]], Dataset]
+EvalData = _EvalDataObject[Input, Expected] | type[_EvalDataObject[Input, Expected]] | Dataset
 
-EvalTask = Union[
-    Callable[[Input], Union[Output, Awaitable[Output]]],
-    Callable[[Input, EvalHooks[Expected]], Union[Output, Awaitable[Output]]],
-]
+EvalTask = (
+    Callable[[Input], Output | Awaitable[Output]] | Callable[[Input, EvalHooks[Expected]], Output | Awaitable[Output]]
+)
 
 ErrorScoreHandler = Callable[[Span, EvalCase[Input, Expected], Sequence[str]], dict[str, float] | None]
 

py/src/braintrust/functions/stream.py

@@ -9,7 +9,7 @@
 import json
 from collections.abc import Generator, Iterable
 from itertools import tee
-from typing import Literal, Union
+from typing import Literal
 
 from sseclient import SSEClient
 
@@ -79,13 +79,9 @@ class BraintrustInvokeError(ValueError):
     pass
 
 
-BraintrustStreamChunk = Union[
-    BraintrustTextChunk,
-    BraintrustJsonChunk,
-    BraintrustErrorChunk,
-    BraintrustConsoleChunk,
-    BraintrustProgressChunk,
-]
+BraintrustStreamChunk = (
+    BraintrustTextChunk | BraintrustJsonChunk | BraintrustErrorChunk | BraintrustConsoleChunk | BraintrustProgressChunk
+)
 
 
 class BraintrustStream:

py/src/braintrust/integrations/claude_agent_sdk/_constants.py

@@ -1,7 +1,8 @@
+from collections.abc import Mapping
 from dataclasses import dataclass
 from enum import Enum
 from types import MappingProxyType
-from typing import Final, Mapping
+from typing import Final
 
 
 class MessageClassName(str, Enum):

py/src/braintrust/integrations/langchain/callbacks.py

@@ -7,7 +7,6 @@
 from typing import (
     Any,
     TypedDict,
-    Union,
 )
 from uuid import UUID
 
@@ -528,7 +527,7 @@ def on_llm_new_token(
         self,
         token: str,
         *,
-        chunk: Union["GenerationChunk", "ChatGenerationChunk"] | None = None,  # type: ignore
+        chunk: "GenerationChunk | ChatGenerationChunk | None" = None,  # type: ignore
         run_id: UUID,
         parent_run_id: UUID | None = None,
         **kwargs: Any,

py/src/braintrust/integrations/langchain/helpers.py

@@ -1,4 +1,5 @@
-from typing import Any, Sequence
+from collections.abc import Sequence
+from typing import Any
 from unittest.mock import ANY
 
 

py/src/braintrust/integrations/langchain/test_callbacks.py

@@ -3,7 +3,7 @@
 import time
 import uuid
 from pathlib import Path
-from typing import Dict, List, Union, cast
+from typing import cast
 
 import pytest
 from braintrust import logger
@@ -56,7 +56,7 @@ def test_llm_calls(logger_memory_logger):
         presence_penalty=0,
         n=1,
     )
-    chain: RunnableSerializable[Dict[str, str], BaseMessage] = prompt.pipe(model)
+    chain: RunnableSerializable[dict[str, str], BaseMessage] = prompt.pipe(model)
     chain.invoke({"number": "2"}, config={"callbacks": [cast(BaseCallbackHandler, handler)]})
 
     spans = memory_logger.pop()
@@ -159,7 +159,7 @@ def test_chain_with_memory(logger_memory_logger):
     handler = BraintrustCallbackHandler(logger=test_logger)
     prompt = ChatPromptTemplate.from_template("{history} User: {input}")
     model = ChatOpenAI(model="gpt-4o-mini")
-    chain: RunnableSerializable[Dict[str, str], BaseMessage] = prompt.pipe(model)
+    chain: RunnableSerializable[dict[str, str], BaseMessage] = prompt.pipe(model)
 
     memory = {"history": "Assistant: Hello! How can I assist you today?"}
     chain.invoke(
@@ -399,7 +399,7 @@ def test_parallel_execution(logger_memory_logger):
 
     map_chain.invoke({"topic": "bear"}, config={"callbacks": [cast(BaseCallbackHandler, handler)]})
 
-    spans = cast(List, memory_logger.pop())
+    spans = cast(list, memory_logger.pop())
 
     # Find the LLM spans
     llm_spans = find_spans_by_attributes(spans, name="ChatOpenAI")
@@ -480,16 +480,16 @@ def test_langgraph_state_management(logger_memory_logger):
         n=1,
     )
 
-    def say_hello(state: Dict[str, str]):
+    def say_hello(state: dict[str, str]):
         response = model.invoke("Say hello")
-        return cast(Union[str, List[str], Dict[str, str]], response.content)
+        return cast(str | list[str] | dict[str, str], response.content)
 
-    def say_bye(state: Dict[str, str]):
+    def say_bye(state: dict[str, str]):
         print("From the 'sayBye' node: Bye world!")
         return "Bye"
 
     workflow = (
-        StateGraph(state_schema=Dict[str, str])
+        StateGraph(state_schema=dict[str, str])
         .add_node("sayHello", say_hello)
         .add_node("sayBye", say_bye)
         .add_edge(START, "sayHello")
@@ -837,10 +837,10 @@ def test_streaming_ttft(logger_memory_logger):
         max_completion_tokens=50,
         streaming=True,
     )
-    chain: RunnableSerializable[Dict[str, str], BaseMessage] = prompt.pipe(model)
+    chain: RunnableSerializable[dict[str, str], BaseMessage] = prompt.pipe(model)
 
     # Collect chunks to verify streaming works
-    chunks: List[str] = []
+    chunks: list[str] = []
     for chunk in chain.stream({}, config={"callbacks": [cast(BaseCallbackHandler, handler)]}):
         if chunk.content:
             chunks.append(str(chunk.content))
@@ -1272,9 +1272,9 @@ async def test_async_streaming(logger_memory_logger):
     handler = BraintrustCallbackHandler(logger=test_logger)
     prompt = ChatPromptTemplate.from_template("Count from 1 to 3.")
     model = ChatOpenAI(model="gpt-4o-mini", max_completion_tokens=20, streaming=True)
-    chain: RunnableSerializable[Dict[str, str], BaseMessage] = prompt.pipe(model)
+    chain: RunnableSerializable[dict[str, str], BaseMessage] = prompt.pipe(model)
 
-    chunks: List[str] = []
+    chunks: list[str] = []
     async for chunk in chain.astream({}, config={"callbacks": [cast(BaseCallbackHandler, handler)]}):
         if chunk.content:
             chunks.append(str(chunk.content))

py/src/braintrust/integrations/langchain/test_context.py

@@ -1,5 +1,4 @@
 # pyright: reportTypedDictNotRequiredAccess=none
-from typing import Dict
 from unittest.mock import ANY
 
 import pytest
@@ -57,7 +56,7 @@ def test_global_handler(logger_memory_logger):
         presence_penalty=0,
         n=1,
     )
-    chain: RunnableSerializable[Dict[str, str], BaseMessage] = prompt.pipe(model)
+    chain: RunnableSerializable[dict[str, str], BaseMessage] = prompt.pipe(model)
 
     message = chain.invoke({"number": "2"})
 

py/src/braintrust/integrations/pydantic_ai/test_pydantic_ai_wrap_openai.py

@@ -1,5 +1,5 @@
 import time
-from typing import Any, Dict
+from typing import Any
 
 import pytest
 from openai import AsyncOpenAI
@@ -61,7 +61,7 @@ def memory_logger():
         yield bgl
 
 
-def _assert_metrics_are_valid(metrics: Dict[str, Any]):
+def _assert_metrics_are_valid(metrics: dict[str, Any]):
     assert metrics["tokens"] > 0
     assert metrics["prompt_tokens"] > 0
     assert metrics["completion_tokens"] > 0

py/src/braintrust/logger.py

@@ -28,7 +28,6 @@
     Literal,
     TypedDict,
     TypeVar,
-    Union,
     cast,
     overload,
 )
@@ -1600,7 +1599,7 @@ def init(
     base_experiment_id: str | None = None,
     repo_info: RepoInfo | None = None,
     state: BraintrustState | None = None,
-) -> Union["Experiment", "ReadonlyExperiment"]:
+) -> "Experiment | ReadonlyExperiment":
     """
     Log in, and then initialize a new experiment in a specified project. If the project does not exist, it will be created.
 
@@ -1767,7 +1766,7 @@ def compute_metadata():
     return ret
 
 
-def init_experiment(*args, **kwargs) -> Union["Experiment", "ReadonlyExperiment"]:
+def init_experiment(*args, **kwargs) -> "Experiment | ReadonlyExperiment":
     """Alias for `init`"""
 
     return init(*args, **kwargs)
@@ -2392,7 +2391,7 @@ def parent_context(parent: str | None, state: BraintrustState | None = None):
 
 def get_span_parent_object(
     parent: str | None = None, state: BraintrustState | None = None
-) -> Union[SpanComponentsV4, "Logger", "Experiment", Span]:
+) -> "SpanComponentsV4 | Logger | Experiment | Span":
     """Mainly for internal use. Return the parent object for starting a span in a global context.
     Applies precedence: current span > propagated parent string > experiment > logger."""
 
@@ -4857,7 +4856,7 @@ def __exit__(self, exc_type, exc_value, traceback) -> None:
 def render_message(render: Callable[[str], str], message: PromptMessage):
     base = {k: v for (k, v) in message.as_dict().items() if v is not None}
     # TODO: shouldn't load_prompt guarantee content is a PromptMessage?
-    content = cast(Union[str, list[Union[TextPart, ImagePart]], dict[str, Any]], message.content)
+    content = cast(str | list[TextPart | ImagePart] | dict[str, Any], message.content)
     if content is not None:
         if isinstance(content, str):
             base["content"] = render(content)
@@ -5618,9 +5617,7 @@ def __str__(self):
 
 
 class TracedThreadPoolExecutor(concurrent.futures.ThreadPoolExecutor):
-    # Returns Any because Future[T] generic typing was stabilized in Python 3.9,
-    # but we maintain compatibility with older type checkers.
-    def submit(self, fn: Callable[..., Any], *args: Any, **kwargs: Any) -> Any:
+    def submit(self, fn: Callable[..., Any], *args: Any, **kwargs: Any) -> concurrent.futures.Future[Any]:
         # Capture all current context variables
         context = contextvars.copy_context()