Python: [BREAKING] Python: Provider-leading client design & OpenAI package extraction (#4818)

* Python: Provider-leading client design & OpenAI package extraction

Major refactoring of the Python Agent Framework client architecture:

- Extract OpenAI clients into new `agent-framework-openai` package
- Core package no longer depends on openai, azure-identity, azure-ai-projects
- Rename clients for discoverability: OpenAIResponsesClient → OpenAIChatClient,
  OpenAIChatClient → OpenAIChatCompletionClient
- Unify `model_id`/`deployment_name`/`model_deployment_name` → `model` param
- New FoundryChatClient for Azure AI Foundry Responses API
- New FoundryAgent/FoundryAgentClient for connecting to pre-configured Foundry agents
- Remove OpenAIBase/OpenAIConfigMixin from non-deprecated client MRO
- Deprecate AzureOpenAI* clients, AzureAIClient, OpenAIAssistantsClient
- Reorganize samples: azure_openai+azure_ai+azure_ai_agent → azure/
- ADR-0020: Provider-Leading Client Design

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* fix: missing Agent imports in samples, .model_id → .model in foundry_local sample

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* fix: CI failures — mypy errors, coverage targets, sample imports

- azure-ai mypy: add type ignores for TypedDict total=, model arg, forward ref
- Coverage: replace core.azure/openai targets with openai package target
- project_provider: add type annotation for opts dict

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* fix: populate openai .pyi stub, fix broken README links, coverage targets

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* fixes

* updated observabilitty

* reset azure init.pyi

* fix errors

* updated adr number

* fix foundry local

* fixed not renamed docstrings and comments, and added deprecated markers to old classes

* fix tests and pyprojects

* fix test vars

* updated function tests

* update durable

* updated test setup for functions

* Fix Foundry auth in workflow samples

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Stabilize Python integration workflows

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Update hosting samples for Foundry

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Trigger full CI rerun

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Trigger CI rerun again

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* trigger rerun

* trigger rerun

* fix for litellm

* undo durabletask changes

* Move Foundry APIs into foundry namespace

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Fix Foundry pyproject formatting

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Split provider samples by Foundry surface

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Restore hosting sample requirements

Also fix the Foundry Local sample link after the provider sample move.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* updated tests

* udpated foundry integration tests

* removed dist from azurefunctions tests

* Use separate Foundry clients for concurrent agents

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* fix client setup in azfunc and durable

* disabled two tests

* updated setup for some function and durable tests

* improved azure openai setup with new clients

* ignore deprecated

* fixes

* skip 11

* remove openai assistants int tests

---------

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
This commit is contained in:
Eduard van Valkenburg
2026-03-25 10:56:29 +01:00
committed by GitHub
Unverified
parent 4b533608b6
commit 5e056b672e
485 changed files with 9784 additions and 12084 deletions
@@ -2,23 +2,35 @@
import importlib.metadata
from ._agent_provider import AzureAIAgentsProvider
from ._chat_client import AzureAIAgentClient, AzureAIAgentOptions
from ._client import AzureAIClient, AzureAIProjectAgentOptions, RawAzureAIClient
from ._agent_provider import AzureAIAgentsProvider # pyright: ignore[reportDeprecated]
from ._chat_client import AzureAIAgentClient, AzureAIAgentOptions # pyright: ignore[reportDeprecated]
from ._client import AzureAIClient, AzureAIProjectAgentOptions, RawAzureAIClient # pyright: ignore[reportDeprecated]
from ._deprecated_azure_openai import (
AzureOpenAIAssistantsClient, # pyright: ignore[reportDeprecated]
AzureOpenAIAssistantsOptions,
AzureOpenAIChatClient, # pyright: ignore[reportDeprecated]
AzureOpenAIChatOptions,
AzureOpenAIConfigMixin,
AzureOpenAIEmbeddingClient, # pyright: ignore[reportDeprecated]
AzureOpenAIResponsesClient, # pyright: ignore[reportDeprecated]
AzureOpenAIResponsesOptions,
AzureOpenAISettings,
AzureUserSecurityContext,
)
from ._embedding_client import (
AzureAIInferenceEmbeddingClient,
AzureAIInferenceEmbeddingOptions,
AzureAIInferenceEmbeddingSettings,
RawAzureAIInferenceEmbeddingClient,
)
from ._foundry_memory_provider import FoundryMemoryProvider
from ._project_provider import AzureAIProjectAgentProvider
from ._entra_id_authentication import AzureCredentialTypes, AzureTokenProvider
from ._project_provider import AzureAIProjectAgentProvider # pyright: ignore[reportDeprecated]
from ._shared import AzureAISettings
try:
__version__ = importlib.metadata.version(__name__)
except importlib.metadata.PackageNotFoundError:
__version__ = "0.0.0" # Fallback for development mode
__version__ = "0.0.0"
__all__ = [
"AzureAIAgentClient",
@@ -31,7 +43,18 @@ __all__ = [
"AzureAIProjectAgentOptions",
"AzureAIProjectAgentProvider",
"AzureAISettings",
"FoundryMemoryProvider",
"AzureCredentialTypes",
"AzureOpenAIAssistantsClient",
"AzureOpenAIAssistantsOptions",
"AzureOpenAIChatClient",
"AzureOpenAIChatOptions",
"AzureOpenAIConfigMixin",
"AzureOpenAIEmbeddingClient",
"AzureOpenAIResponsesClient",
"AzureOpenAIResponsesOptions",
"AzureOpenAISettings",
"AzureTokenProvider",
"AzureUserSecurityContext",
"RawAzureAIClient",
"RawAzureAIInferenceEmbeddingClient",
"__version__",
@@ -18,19 +18,23 @@ from agent_framework import (
from agent_framework._mcp import MCPTool
from agent_framework._settings import load_settings
from agent_framework._tools import ToolTypes
from agent_framework.azure._entra_id_authentication import AzureCredentialTypes
from azure.ai.agents.aio import AgentsClient
from azure.ai.agents.models import Agent as AzureAgent
from azure.ai.agents.models import ResponseFormatJsonSchema, ResponseFormatJsonSchemaType
from pydantic import BaseModel
from ._chat_client import AzureAIAgentClient, AzureAIAgentOptions
from ._chat_client import AzureAIAgentClient, AzureAIAgentOptions # pyright: ignore[reportDeprecated]
from ._entra_id_authentication import AzureCredentialTypes
from ._shared import AzureAISettings, to_azure_ai_agent_tools
if sys.version_info >= (3, 13):
from typing import Self, TypeVar # type: ignore # pragma: no cover
else:
from typing_extensions import Self, TypeVar # type: ignore # pragma: no cover
if sys.version_info >= (3, 13):
from warnings import deprecated # type: ignore # pragma: no cover
else:
from typing_extensions import deprecated # type: ignore # pragma: no cover
if sys.version_info >= (3, 11):
from typing import TypedDict # type: ignore # pragma: no cover
else:
@@ -47,6 +51,11 @@ OptionsCoT = TypeVar(
)
@deprecated(
"AzureAIAgentClient and the AzureAIAgentsProvider are deprecated. "
"They target the V1 Agents Service API and have no direct replacement; "
"for new Foundry projects, use FoundryAgent."
)
class AzureAIAgentsProvider(Generic[OptionsCoT]):
"""Provider for Azure AI Agent Service V1 (Persistent Agents API).
@@ -426,7 +435,7 @@ class AzureAIAgentsProvider(Generic[OptionsCoT]):
context_providers: Context providers to include during agent invocation.
"""
# Create the underlying client
client = AzureAIAgentClient(
client = AzureAIAgentClient( # pyright: ignore[reportDeprecated]
agents_client=self._agents_client,
agent_id=agent.id,
agent_name=agent.name,
@@ -36,7 +36,6 @@ from agent_framework import (
)
from agent_framework._settings import load_settings
from agent_framework._tools import ToolTypes
from agent_framework.azure._entra_id_authentication import AzureCredentialTypes
from agent_framework.exceptions import (
ChatClientException,
ChatClientInvalidRequestException,
@@ -92,12 +91,14 @@ from azure.ai.agents.models import (
)
from pydantic import BaseModel
from ._entra_id_authentication import AzureCredentialTypes
from ._shared import AzureAISettings, resolve_file_ids, to_azure_ai_agent_tools
if sys.version_info >= (3, 13):
from typing import TypeVar # type: ignore # pragma: no cover
from warnings import deprecated # type: ignore # pragma: no cover
else:
from typing_extensions import TypeVar # type: ignore # pragma: no cover
from typing_extensions import TypeVar, deprecated # type: ignore # pragma: no cover
if sys.version_info >= (3, 12):
from typing import override # type: ignore # pragma: no cover
else:
@@ -210,6 +211,11 @@ AzureAIAgentOptionsT = TypeVar(
# endregion
@deprecated(
"AzureAIAgentClient is deprecated. "
"It targets the V1 Agents Service API and has no direct replacement; "
"for new Foundry projects, use FoundryAgent."
)
class AzureAIAgentClient(
FunctionInvocationLayer[AzureAIAgentOptionsT],
ChatMiddlewareLayer[AzureAIAgentOptionsT],
@@ -221,7 +227,8 @@ class AzureAIAgentClient(
.. deprecated::
AzureAIAgentClient is deprecated and will be removed in a future release.
Use :class:`AzureAIClient` instead for the V2 (Projects/Responses) API.
It targets the V1 Agents Service API and has no direct replacement.
For new Foundry projects, use :class:`FoundryAgent`.
"""
OTEL_PROVIDER_NAME: ClassVar[str] = "azure.ai" # type: ignore[reportIncompatibleVariableOverride, misc]
@@ -239,7 +246,8 @@ class AzureAIAgentClient(
.. deprecated::
This method is deprecated and will be removed in a future release.
Use :meth:`AzureAIClient.get_code_interpreter_tool` instead.
For new Foundry projects, configure hosted tools on the Foundry agent definition
in the service instead.
Keyword Args:
file_ids: List of uploaded file IDs or Content objects to make available to
@@ -272,7 +280,7 @@ class AzureAIAgentClient(
"""
warnings.warn(
"AzureAIAgentClient.get_code_interpreter_tool() is deprecated and will be removed in a future release; "
"use AzureAIClient.get_code_interpreter_tool() instead.",
"for new Foundry projects, configure hosted tools on the Foundry agent definition in the service instead.",
DeprecationWarning,
stacklevel=2,
)
@@ -288,7 +296,8 @@ class AzureAIAgentClient(
.. deprecated::
This method is deprecated and will be removed in a future release.
Use :meth:`AzureAIClient.get_file_search_tool` instead.
For new Foundry projects, configure hosted tools on the Foundry agent definition
in the service instead.
Keyword Args:
vector_store_ids: List of vector store IDs to search within.
@@ -308,7 +317,7 @@ class AzureAIAgentClient(
"""
warnings.warn(
"AzureAIAgentClient.get_file_search_tool() is deprecated and will be removed in a future release; "
"use AzureAIClient.get_file_search_tool() instead.",
"for new Foundry projects, configure hosted tools on the Foundry agent definition in the service instead.",
DeprecationWarning,
stacklevel=2,
)
@@ -325,7 +334,8 @@ class AzureAIAgentClient(
.. deprecated::
This method is deprecated and will be removed in a future release.
Use :meth:`AzureAIClient.get_web_search_tool` instead.
For new Foundry projects, configure hosted tools on the Foundry agent definition
in the service instead.
For Azure AI Agents, web search uses Bing Grounding or Bing Custom Search.
If no arguments are provided, attempts to read from environment variables.
@@ -369,7 +379,7 @@ class AzureAIAgentClient(
"""
warnings.warn(
"AzureAIAgentClient.get_web_search_tool() is deprecated and will be removed in a future release; "
"use AzureAIClient.get_web_search_tool() instead.",
"for new Foundry projects, configure hosted tools on the Foundry agent definition in the service instead.",
DeprecationWarning,
stacklevel=2,
)
@@ -410,7 +420,8 @@ class AzureAIAgentClient(
.. deprecated::
This method is deprecated and will be removed in a future release.
Use :meth:`AzureAIClient.get_mcp_tool` instead.
For new Foundry projects, configure hosted tools on the Foundry agent definition
in the service instead.
This configures an MCP (Model Context Protocol) server that will be called
by Azure AI's service. The tools from this MCP server are executed remotely
@@ -446,7 +457,7 @@ class AzureAIAgentClient(
"""
warnings.warn(
"AzureAIAgentClient.get_mcp_tool() is deprecated and will be removed in a future release; "
"use AzureAIClient.get_mcp_tool() instead.",
"for new Foundry projects, configure hosted tools on the Foundry agent definition in the service instead.",
DeprecationWarning,
stacklevel=2,
)
@@ -561,12 +572,6 @@ class AzureAIAgentClient(
client: AzureAIAgentClient[MyOptions] = AzureAIAgentClient(credential=credential)
response = await client.get_response("Hello", options={"my_custom_option": "value"})
"""
warnings.warn(
"AzureAIAgentClient is deprecated and will be removed in a future release; "
"use AzureAIClient instead for the V2 (Projects/Responses) API.",
DeprecationWarning,
stacklevel=2,
)
azure_ai_settings = load_settings(
AzureAISettings,
env_prefix="AZURE_AI_",
@@ -30,10 +30,9 @@ from agent_framework import (
)
from agent_framework._settings import load_settings
from agent_framework._tools import ToolTypes
from agent_framework.azure._entra_id_authentication import AzureCredentialTypes
from agent_framework.observability import ChatTelemetryLayer
from agent_framework.openai import OpenAIResponsesOptions
from agent_framework.openai._responses_client import RawOpenAIResponsesClient
from agent_framework_openai._chat_client import RawOpenAIChatClient
from azure.ai.projects.aio import AIProjectClient
from azure.ai.projects.models import (
ApproximateLocation,
@@ -50,12 +49,14 @@ from azure.ai.projects.models import (
from azure.ai.projects.models import FileSearchTool as ProjectsFileSearchTool
from azure.core.exceptions import ResourceNotFoundError
from ._entra_id_authentication import AzureCredentialTypes
from ._shared import AzureAISettings, create_text_format_config, resolve_file_ids
if sys.version_info >= (3, 13):
from typing import TypeVar # type: ignore # pragma: no cover
from warnings import deprecated # type: ignore # pragma: no cover
else:
from typing_extensions import TypeVar # type: ignore # pragma: no cover
from typing_extensions import TypeVar, deprecated # type: ignore # pragma: no cover
if sys.version_info >= (3, 12):
from typing import override # type: ignore # pragma: no cover
else:
@@ -68,7 +69,7 @@ else:
logger = logging.getLogger("agent_framework.azure")
class AzureAIProjectAgentOptions(OpenAIResponsesOptions, total=False):
class AzureAIProjectAgentOptions(OpenAIResponsesOptions, total=False): # type: ignore[misc, call-arg]
"""Azure AI Project Agent options."""
rai_config: RaiConfig
@@ -88,8 +89,13 @@ AzureAIClientOptionsT = TypeVar(
_DOC_INDEX_PATTERN = re.compile(r"doc_(\d+)")
class RawAzureAIClient(RawOpenAIResponsesClient[AzureAIClientOptionsT], Generic[AzureAIClientOptionsT]):
"""Raw Azure AI client without middleware, telemetry, or function invocation layers.
@deprecated(
"RawAzureAIClient is deprecated. "
"Use RawFoundryAgentChatClient for low-level Foundry agent client customization, "
"or FoundryAgent for the recommended production API."
)
class RawAzureAIClient(RawOpenAIChatClient[AzureAIClientOptionsT], Generic[AzureAIClientOptionsT]):
"""Deprecated raw Azure AI client without middleware, telemetry, or function invocation layers.
Warning:
**This class should not normally be used directly.** It does not include middleware,
@@ -101,7 +107,8 @@ class RawAzureAIClient(RawOpenAIResponsesClient[AzureAIClientOptionsT], Generic[
2. **ChatMiddlewareLayer** - Applies chat middleware per model call and stays outside telemetry
3. **ChatTelemetryLayer** - Must stay inside chat middleware for correct per-call telemetry
Use ``AzureAIClient`` instead for a fully-featured client with all layers applied.
Use ``RawFoundryAgentChatClient`` for low-level Foundry agent customization, or
``FoundryAgent`` for the recommended production API.
"""
OTEL_PROVIDER_NAME: ClassVar[str] = "azure.ai" # type: ignore[reportIncompatibleVariableOverride, misc]
@@ -215,8 +222,10 @@ class RawAzureAIClient(RawOpenAIResponsesClient[AzureAIClientOptionsT], Generic[
project_client = AIProjectClient(**project_client_kwargs)
should_close_client = True
# Initialize parent
super().__init__(
# Initialize parent with OpenAI client from project
super().__init__( # type: ignore
async_client=project_client.get_openai_client(),
model=azure_ai_settings.get("model"), # type: ignore[arg-type]
additional_properties=additional_properties,
)
@@ -680,10 +689,6 @@ class RawAzureAIClient(RawOpenAIResponsesClient[AzureAIClientOptionsT], Generic[
return result, instructions
async def _initialize_client(self) -> None:
"""Initialize OpenAI client."""
self.client = self.project_client.get_openai_client() # type: ignore
def _update_agent_name_and_description(self, agent_name: str | None, description: str | None = None) -> None:
"""Update the agent name in the chat client.
@@ -842,7 +847,7 @@ class RawAzureAIClient(RawOpenAIResponsesClient[AzureAIClientOptionsT], Generic[
if not stream:
async def _enrich_response() -> ChatResponse:
response = await super(RawAzureAIClient, self)._inner_get_response(
response = await super(RawAzureAIClient, self)._inner_get_response( # pyright: ignore[reportDeprecated]
messages=messages, options=options, stream=False, **kwargs
)
get_urls = self._extract_azure_search_urls(response.raw_representation.output) # type: ignore[union-attr]
@@ -1182,8 +1187,8 @@ class RawAzureAIClient(RawOpenAIResponsesClient[AzureAIClientOptionsT], Generic[
It does NOT create an agent on the Azure AI service - the actual agent
will be created on the server during the first invocation (run).
For creating and managing persistent agents on the server, use
:class:`~agent_framework_azure_ai.AzureAIProjectAgentProvider` instead.
For working with pre-configured persistent agents on the server, use
:class:`~agent_framework_azure_ai.FoundryAgent` instead.
Keyword Args:
id: The unique identifier for the agent. Will be created automatically if not provided.
@@ -1213,21 +1218,23 @@ class RawAzureAIClient(RawOpenAIResponsesClient[AzureAIClientOptionsT], Generic[
)
@deprecated("AzureAIClient is deprecated. Use FoundryAgent instead.")
class AzureAIClient(
FunctionInvocationLayer[AzureAIClientOptionsT],
ChatMiddlewareLayer[AzureAIClientOptionsT],
ChatTelemetryLayer[AzureAIClientOptionsT],
RawAzureAIClient[AzureAIClientOptionsT],
RawAzureAIClient[AzureAIClientOptionsT], # pyright: ignore[reportDeprecated]
Generic[AzureAIClientOptionsT],
):
"""Azure AI client with middleware, telemetry, and function invocation support.
"""Deprecated Azure AI client with middleware, telemetry, and function invocation support.
This is the recommended client for most use cases. It includes:
This class is deprecated. Use ``FoundryAgent`` instead for connecting to
pre-configured agents in Foundry. It includes:
- Chat middleware support for request/response interception
- OpenTelemetry-based telemetry for observability
- Automatic function/tool invocation handling
For a minimal implementation without these features, use :class:`RawAzureAIClient`.
For a minimal implementation without these features, use :class:`RawFoundryAgentChatClient`.
"""
def __init__(
@@ -0,0 +1,897 @@
# Copyright (c) Microsoft. All rights reserved.
"""Deprecated Azure OpenAI client classes.
All classes in this module are deprecated and will be removed in a future release.
Migrate to the ``agent_framework_openai`` package equivalents with an ``AsyncAzureOpenAI`` client,
or use ``FoundryChatClient`` for Azure AI Foundry projects.
"""
from __future__ import annotations
import json
import logging
import sys
from collections.abc import Mapping, Sequence
from copy import copy
from typing import TYPE_CHECKING, Any, ClassVar, Final, Generic, cast
from urllib.parse import urljoin, urlparse
from agent_framework._middleware import ChatMiddlewareLayer
from agent_framework._settings import SecretString, load_settings
from agent_framework._telemetry import AGENT_FRAMEWORK_USER_AGENT, APP_INFO, prepend_agent_framework_to_user_agent
from agent_framework._tools import FunctionInvocationConfiguration, FunctionInvocationLayer
from agent_framework._types import Annotation, Content
from agent_framework.observability import ChatTelemetryLayer, EmbeddingTelemetryLayer
from agent_framework_openai._assistants_client import OpenAIAssistantsClient, OpenAIAssistantsOptions
from agent_framework_openai._chat_client import OpenAIChatOptions, RawOpenAIChatClient
from agent_framework_openai._chat_completion_client import OpenAIChatCompletionOptions, RawOpenAIChatCompletionClient
from agent_framework_openai._embedding_client import OpenAIEmbeddingOptions, RawOpenAIEmbeddingClient
from agent_framework_openai._shared import OpenAIBase
from azure.ai.projects.aio import AIProjectClient
from openai import AsyncOpenAI
from openai.lib.azure import AsyncAzureOpenAI
from pydantic import BaseModel
from ._entra_id_authentication import AzureCredentialTypes, AzureTokenProvider, resolve_credential_to_token_provider
if sys.version_info >= (3, 13):
from typing import TypeVar # type: ignore # pragma: no cover
from warnings import deprecated # type: ignore # pragma: no cover
else:
from typing_extensions import TypeVar, deprecated # type: ignore # pragma: no cover
if sys.version_info >= (3, 12):
from typing import override # type: ignore # pragma: no cover
else:
from typing_extensions import override # type: ignore # pragma: no cover
if sys.version_info >= (3, 11):
from typing import TypedDict # type: ignore # pragma: no cover
else:
from typing_extensions import TypedDict # type: ignore # pragma: no cover
if TYPE_CHECKING:
from agent_framework._middleware import MiddlewareTypes
from openai.types.chat.chat_completion import Choice
from openai.types.chat.chat_completion_chunk import Choice as ChunkChoice
logger: logging.Logger = logging.getLogger(__name__)
# region Constants and Settings
DEFAULT_AZURE_API_VERSION: Final[str] = "2024-10-21"
DEFAULT_AZURE_TOKEN_ENDPOINT: Final[str] = "https://cognitiveservices.azure.com/.default" # noqa: S105
class AzureOpenAISettings(TypedDict, total=False):
"""AzureOpenAI model settings.
Settings are resolved in this order: explicit keyword arguments, values from an
explicitly provided .env file, then environment variables with the prefix
'AZURE_OPENAI_'. If settings are missing after resolution, validation will fail.
Keyword Args:
endpoint: The endpoint of the Azure deployment.
chat_deployment_name: The name of the Azure Chat deployment.
responses_deployment_name: The name of the Azure Responses deployment.
embedding_deployment_name: The name of the Azure Embedding deployment.
api_key: The API key for the Azure deployment.
api_version: The API version to use.
base_url: The url of the Azure deployment.
token_endpoint: The token endpoint to use to retrieve the authentication token.
"""
chat_deployment_name: str | None
responses_deployment_name: str | None
embedding_deployment_name: str | None
endpoint: str | None
base_url: str | None
api_key: SecretString | None
api_version: str | None
token_endpoint: str | None
def _apply_azure_defaults(
settings: AzureOpenAISettings,
default_api_version: str = DEFAULT_AZURE_API_VERSION,
default_token_endpoint: str = DEFAULT_AZURE_TOKEN_ENDPOINT,
) -> None:
"""Apply default values for api_version and token_endpoint after loading settings.
Args:
settings: The loaded Azure OpenAI settings dict.
default_api_version: The default API version to use if not set.
default_token_endpoint: The default token endpoint to use if not set.
"""
if not settings.get("api_version"):
settings["api_version"] = default_api_version
if not settings.get("token_endpoint"):
settings["token_endpoint"] = default_token_endpoint
# endregion
# region AzureOpenAIConfigMixin
class AzureOpenAIConfigMixin(OpenAIBase):
"""Internal class for configuring a connection to an Azure OpenAI service."""
OTEL_PROVIDER_NAME: ClassVar[str] = "azure.ai.openai"
def __init__(
self,
deployment_name: str,
endpoint: str | None = None,
base_url: str | None = None,
api_version: str = DEFAULT_AZURE_API_VERSION,
api_key: str | None = None,
token_endpoint: str | None = None,
credential: AzureCredentialTypes | AzureTokenProvider | None = None,
default_headers: Mapping[str, str] | None = None,
client: AsyncOpenAI | None = None,
instruction_role: str | None = None,
**kwargs: Any,
) -> None:
"""Configure a connection to an Azure OpenAI service.
Args:
deployment_name: Name of the deployment.
endpoint: The specific endpoint URL for the deployment.
base_url: The base URL for Azure services.
api_version: Azure API version.
api_key: API key for Azure services.
token_endpoint: Azure AD token scope.
credential: Azure credential or token provider for authentication.
default_headers: Default headers for HTTP requests.
client: An existing client to use.
instruction_role: The role to use for 'instruction' messages.
kwargs: Additional keyword arguments.
"""
merged_headers = dict(copy(default_headers)) if default_headers else {}
if APP_INFO:
merged_headers.update(APP_INFO)
merged_headers = prepend_agent_framework_to_user_agent(merged_headers)
if not client:
ad_token_provider = None
if not api_key and credential:
ad_token_provider = resolve_credential_to_token_provider(credential, token_endpoint)
if not api_key and not ad_token_provider:
raise ValueError("Please provide either api_key, credential, or a client.")
if not endpoint and not base_url:
raise ValueError("Please provide an endpoint or a base_url")
args: dict[str, Any] = {
"default_headers": merged_headers,
}
if api_version:
args["api_version"] = api_version
if ad_token_provider:
args["azure_ad_token_provider"] = ad_token_provider
if api_key:
args["api_key"] = api_key
if base_url:
args["base_url"] = str(base_url)
if endpoint and not base_url:
args["azure_endpoint"] = str(endpoint)
if deployment_name:
args["azure_deployment"] = deployment_name
if "websocket_base_url" in kwargs:
args["websocket_base_url"] = kwargs.pop("websocket_base_url")
client = AsyncAzureOpenAI(**args)
self.endpoint = str(endpoint)
self.base_url = str(base_url)
self.api_version = api_version
self.deployment_name = deployment_name
self.instruction_role = instruction_role
if default_headers:
from agent_framework._telemetry import USER_AGENT_KEY
def_headers = {k: v for k, v in default_headers.items() if k != USER_AGENT_KEY}
else:
def_headers = None
self.default_headers = def_headers
super().__init__(model_id=deployment_name, client=client, **kwargs)
# endregion
# region AzureOpenAIResponsesClient
AzureOpenAIResponsesOptionsT = TypeVar(
"AzureOpenAIResponsesOptionsT",
bound=TypedDict, # type: ignore[valid-type]
default="OpenAIChatOptions",
covariant=True,
)
AzureOpenAIResponsesOptions = OpenAIChatOptions
@deprecated(
"AzureOpenAIResponsesClient is deprecated. "
"Use OpenAIChatClient with an AsyncAzureOpenAI client, or FoundryChatClient for Foundry projects."
)
class AzureOpenAIResponsesClient( # type: ignore[misc]
FunctionInvocationLayer[AzureOpenAIResponsesOptionsT],
ChatMiddlewareLayer[AzureOpenAIResponsesOptionsT],
ChatTelemetryLayer[AzureOpenAIResponsesOptionsT],
RawOpenAIChatClient[AzureOpenAIResponsesOptionsT],
Generic[AzureOpenAIResponsesOptionsT],
):
"""Deprecated Azure Responses client. Use OpenAIChatClient with an AsyncAzureOpenAI client instead."""
OTEL_PROVIDER_NAME: ClassVar[str] = "azure.ai.openai"
def __init__(
self,
*,
api_key: str | None = None,
deployment_name: str | None = None,
endpoint: str | None = None,
base_url: str | None = None,
api_version: str | None = None,
token_endpoint: str | None = None,
credential: AzureCredentialTypes | AzureTokenProvider | None = None,
default_headers: Mapping[str, str] | None = None,
async_client: AsyncOpenAI | None = None,
project_client: Any | None = None,
project_endpoint: str | None = None,
allow_preview: bool | None = None,
env_file_path: str | None = None,
env_file_encoding: str | None = None,
instruction_role: str | None = None,
middleware: Sequence[MiddlewareTypes] | None = None,
function_invocation_configuration: FunctionInvocationConfiguration | None = None,
**kwargs: Any,
) -> None:
"""Initialize an Azure OpenAI Responses client.
Keyword Args:
api_key: The API key.
deployment_name: The deployment name.
endpoint: The deployment endpoint.
base_url: The deployment base URL.
api_version: The deployment API version.
token_endpoint: The token endpoint to request an Azure token.
credential: Azure credential or token provider for authentication.
default_headers: Default headers for HTTP requests.
async_client: An existing client to use.
project_client: An existing AIProjectClient to use.
project_endpoint: The Azure AI Foundry project endpoint URL.
allow_preview: Enables preview opt-in on internally-created AIProjectClient.
env_file_path: Path to .env file for settings.
env_file_encoding: Encoding for .env file.
instruction_role: The role to use for 'instruction' messages.
middleware: Optional sequence of middleware.
function_invocation_configuration: Optional function invocation configuration.
kwargs: Additional keyword arguments.
"""
if (model_id := kwargs.pop("model_id", None)) and not deployment_name:
deployment_name = str(model_id)
if async_client is None and (project_client is not None or project_endpoint is not None):
async_client = self._create_client_from_project(
project_client=project_client,
project_endpoint=project_endpoint,
credential=credential,
allow_preview=allow_preview,
)
azure_openai_settings = load_settings(
AzureOpenAISettings,
env_prefix="AZURE_OPENAI_",
api_key=api_key,
base_url=base_url,
endpoint=endpoint,
responses_deployment_name=deployment_name,
api_version=api_version,
env_file_path=env_file_path,
env_file_encoding=env_file_encoding,
token_endpoint=token_endpoint,
)
_apply_azure_defaults(azure_openai_settings, default_api_version="preview")
endpoint_value = azure_openai_settings.get("endpoint")
if (
not azure_openai_settings.get("base_url")
and endpoint_value
and (hostname := urlparse(str(endpoint_value)).hostname)
and hostname.endswith(".openai.azure.com")
):
azure_openai_settings["base_url"] = urljoin(str(endpoint_value), "/openai/v1/")
responses_deployment_name = azure_openai_settings.get("responses_deployment_name")
if not responses_deployment_name:
raise ValueError(
"Azure OpenAI deployment name is required. Set via 'deployment_name' parameter "
"or 'AZURE_OPENAI_RESPONSES_DEPLOYMENT_NAME' environment variable."
)
if not async_client:
# Create the Azure OpenAI client directly
merged_headers = dict(copy(default_headers)) if default_headers else {}
if APP_INFO:
merged_headers.update(APP_INFO)
merged_headers = prepend_agent_framework_to_user_agent(merged_headers)
api_key_secret = azure_openai_settings.get("api_key")
ad_token_provider = None
if not api_key_secret and credential:
ad_token_provider = resolve_credential_to_token_provider(
credential, azure_openai_settings.get("token_endpoint")
)
if not api_key_secret and not ad_token_provider:
raise ValueError("Please provide either api_key, credential, or a client.")
client_endpoint = azure_openai_settings.get("endpoint")
client_base_url = azure_openai_settings.get("base_url")
if not client_endpoint and not client_base_url:
raise ValueError("Please provide an endpoint or a base_url")
client_args: dict[str, Any] = {"default_headers": merged_headers}
if resolved_api_version := azure_openai_settings.get("api_version"):
client_args["api_version"] = resolved_api_version
if ad_token_provider:
client_args["azure_ad_token_provider"] = ad_token_provider
if api_key_secret:
client_args["api_key"] = api_key_secret.get_secret_value()
if client_base_url:
client_args["base_url"] = str(client_base_url)
if client_endpoint and not client_base_url:
client_args["azure_endpoint"] = str(client_endpoint)
if responses_deployment_name:
client_args["azure_deployment"] = responses_deployment_name
if "websocket_base_url" in kwargs:
client_args["websocket_base_url"] = kwargs.pop("websocket_base_url")
async_client = AsyncAzureOpenAI(**client_args)
# Store Azure-specific attributes for serialization
self.endpoint = str(endpoint_value) if endpoint_value else None
self.api_version = azure_openai_settings.get("api_version") or ""
self.deployment_name = responses_deployment_name
super().__init__(
async_client=async_client,
model=responses_deployment_name,
api_version=azure_openai_settings.get("api_version"),
instruction_role=instruction_role,
default_headers=default_headers,
middleware=middleware, # type: ignore[arg-type]
function_invocation_configuration=function_invocation_configuration,
**kwargs,
)
@staticmethod
def _create_client_from_project(
*,
project_client: AIProjectClient | None,
project_endpoint: str | None,
credential: AzureCredentialTypes | AzureTokenProvider | None,
allow_preview: bool | None = None,
) -> AsyncOpenAI:
"""Create an AsyncOpenAI client from an Azure AI Foundry project."""
if project_client is not None:
return project_client.get_openai_client()
if not project_endpoint:
raise ValueError("Azure AI project endpoint is required when project_client is not provided.")
if not credential:
raise ValueError("Azure credential is required when using project_endpoint without a project_client.")
project_client_kwargs: dict[str, Any] = {
"endpoint": project_endpoint,
"credential": credential, # type: ignore[arg-type]
"user_agent": AGENT_FRAMEWORK_USER_AGENT,
}
if allow_preview is not None:
project_client_kwargs["allow_preview"] = allow_preview
project_client = AIProjectClient(**project_client_kwargs)
return project_client.get_openai_client()
@override
def _check_model_presence(self, options: dict[str, Any]) -> None:
if not options.get("model"):
if not self.model:
raise ValueError("deployment_name must be a non-empty string")
options["model"] = self.model
# endregion
# region AzureOpenAIChatClient
ResponseModelT = TypeVar("ResponseModelT", bound=BaseModel | None, default=None)
class AzureUserSecurityContext(TypedDict, total=False):
"""User security context for Azure AI applications.
These fields help security operations teams investigate and mitigate security
incidents by providing context about the application and end user.
"""
application_name: str
"""Name of the application making the request."""
end_user_id: str
"""Unique identifier for the end user (recommend hashing username/email)."""
end_user_tenant_id: str
"""Microsoft 365 tenant ID the end user belongs to. Required for multi-tenant apps."""
source_ip: str
"""The original client's IP address."""
class AzureOpenAIChatOptions(OpenAIChatCompletionOptions[ResponseModelT], Generic[ResponseModelT], total=False):
"""Azure OpenAI-specific chat options dict.
Extends OpenAIChatCompletionOptions with Azure-specific options including
the "On Your Data" feature and enhanced security context.
"""
data_sources: list[dict[str, Any]]
"""Azure "On Your Data" data sources for retrieval-augmented generation."""
user_security_context: AzureUserSecurityContext
"""Enhanced security context for Azure Defender integration."""
n: int
"""Number of chat completion choices to generate for each input message."""
AzureOpenAIChatOptionsT = TypeVar(
"AzureOpenAIChatOptionsT",
bound=TypedDict, # type: ignore[valid-type]
default="AzureOpenAIChatOptions",
covariant=True,
)
@deprecated("AzureOpenAIChatClient is deprecated. Use OpenAIChatCompletionClient with an AsyncAzureOpenAI client.")
class AzureOpenAIChatClient( # type: ignore[misc]
FunctionInvocationLayer[AzureOpenAIChatOptionsT],
ChatMiddlewareLayer[AzureOpenAIChatOptionsT],
ChatTelemetryLayer[AzureOpenAIChatOptionsT],
RawOpenAIChatCompletionClient[AzureOpenAIChatOptionsT],
Generic[AzureOpenAIChatOptionsT],
):
"""Deprecated Azure OpenAI Chat client. Use OpenAIChatCompletionClient with AsyncAzureOpenAI instead."""
OTEL_PROVIDER_NAME: ClassVar[str] = "azure.ai.openai"
def __init__(
self,
*,
api_key: str | None = None,
deployment_name: str | None = None,
endpoint: str | None = None,
base_url: str | None = None,
api_version: str | None = None,
token_endpoint: str | None = None,
credential: AzureCredentialTypes | AzureTokenProvider | None = None,
default_headers: Mapping[str, str] | None = None,
async_client: AsyncAzureOpenAI | None = None,
additional_properties: dict[str, Any] | None = None,
env_file_path: str | None = None,
env_file_encoding: str | None = None,
instruction_role: str | None = None,
middleware: Sequence[MiddlewareTypes] | None = None,
function_invocation_configuration: FunctionInvocationConfiguration | None = None,
) -> None:
"""Initialize an Azure OpenAI Chat completion client.
Keyword Args:
api_key: The API key.
deployment_name: The deployment name.
endpoint: The deployment endpoint.
base_url: The deployment base URL.
api_version: The deployment API version.
token_endpoint: The token endpoint to request an Azure token.
credential: Azure credential or token provider for authentication.
default_headers: Default headers for HTTP requests.
async_client: An existing client to use.
additional_properties: Additional properties stored on the client instance.
env_file_path: Path to .env file for settings.
env_file_encoding: Encoding for .env file.
instruction_role: The role to use for 'instruction' messages.
middleware: Optional sequence of middleware.
function_invocation_configuration: Optional function invocation configuration.
"""
azure_openai_settings = load_settings(
AzureOpenAISettings,
env_prefix="AZURE_OPENAI_",
api_key=api_key,
base_url=base_url,
endpoint=endpoint,
chat_deployment_name=deployment_name,
api_version=api_version,
env_file_path=env_file_path,
env_file_encoding=env_file_encoding,
token_endpoint=token_endpoint,
)
_apply_azure_defaults(azure_openai_settings)
chat_deployment_name = azure_openai_settings.get("chat_deployment_name")
if not chat_deployment_name:
raise ValueError(
"Azure OpenAI deployment name is required. Set via 'deployment_name' parameter "
"or 'AZURE_OPENAI_CHAT_DEPLOYMENT_NAME' environment variable."
)
if not async_client:
# Create the Azure OpenAI client directly
merged_headers = dict(copy(default_headers)) if default_headers else {}
if APP_INFO:
merged_headers.update(APP_INFO)
merged_headers = prepend_agent_framework_to_user_agent(merged_headers)
api_key_secret = azure_openai_settings.get("api_key")
ad_token_provider = None
if not api_key_secret and credential:
ad_token_provider = resolve_credential_to_token_provider(
credential, azure_openai_settings.get("token_endpoint")
)
if not api_key_secret and not ad_token_provider:
raise ValueError("Please provide either api_key, credential, or a client.")
endpoint_value = azure_openai_settings.get("endpoint")
base_url_value = azure_openai_settings.get("base_url")
if not endpoint_value and not base_url_value:
raise ValueError("Please provide an endpoint or a base_url")
client_args: dict[str, Any] = {"default_headers": merged_headers}
if resolved_api_version := azure_openai_settings.get("api_version"):
client_args["api_version"] = resolved_api_version
if ad_token_provider:
client_args["azure_ad_token_provider"] = ad_token_provider
if api_key_secret:
client_args["api_key"] = api_key_secret.get_secret_value()
if base_url_value:
client_args["base_url"] = str(base_url_value)
if endpoint_value and not base_url_value:
client_args["azure_endpoint"] = str(endpoint_value)
if chat_deployment_name:
client_args["azure_deployment"] = chat_deployment_name
async_client = AsyncAzureOpenAI(**client_args)
# Store Azure-specific attributes for serialization
self.endpoint = str(azure_openai_settings.get("endpoint") or "")
self.api_version = azure_openai_settings.get("api_version") or ""
self.deployment_name = chat_deployment_name
super().__init__(
async_client=async_client,
model=chat_deployment_name,
api_version=azure_openai_settings.get("api_version"),
instruction_role=instruction_role,
default_headers=default_headers,
additional_properties=additional_properties,
middleware=middleware, # type: ignore[arg-type]
function_invocation_configuration=function_invocation_configuration,
)
@override
def _parse_text_from_openai(self, choice: Choice | ChunkChoice) -> Content | None:
"""Parse the choice into a Content object with type='text'.
Overwritten from RawOpenAIChatCompletionClient to deal with Azure On Your Data function.
"""
message = getattr(choice, "message", None)
if message is None:
message = getattr(choice, "delta", None)
if message is None: # type: ignore
return None
if hasattr(message, "refusal") and message.refusal:
return Content.from_text(text=message.refusal, raw_representation=choice)
if not message.content:
return None
text_content = Content.from_text(text=message.content, raw_representation=choice)
if not message.model_extra or "context" not in message.model_extra:
return text_content
context_raw: object = cast(object, message.context) # type: ignore[union-attr]
if isinstance(context_raw, str):
try:
context_raw = json.loads(context_raw)
except json.JSONDecodeError:
logger.warning("Context is not a valid JSON string, ignoring context.")
return text_content
if not isinstance(context_raw, dict):
logger.warning("Context is not a valid dictionary, ignoring context.")
return text_content
context = cast(dict[str, Any], context_raw)
if intent := context.get("intent"):
text_content.additional_properties = {"intent": intent}
citations = context.get("citations")
if isinstance(citations, list) and citations:
annotations: list[Annotation] = []
for citation_raw in cast(list[object], citations):
if not isinstance(citation_raw, dict):
continue
citation = cast(dict[str, Any], citation_raw)
annotations.append(
Annotation(
type="citation",
title=citation.get("title", ""),
url=citation.get("url", ""),
snippet=citation.get("content", ""),
file_id=citation.get("filepath", ""),
tool_name="Azure-on-your-Data",
additional_properties={"chunk_id": citation.get("chunk_id", "")},
raw_representation=citation,
)
)
text_content.annotations = annotations
return text_content
# endregion
# region AzureOpenAIAssistantsClient
AzureOpenAIAssistantsOptionsT = TypeVar(
"AzureOpenAIAssistantsOptionsT",
bound=TypedDict, # type: ignore[valid-type]
default="OpenAIAssistantsOptions",
covariant=True,
)
AzureOpenAIAssistantsOptions = OpenAIAssistantsOptions
@deprecated(
"AzureOpenAIAssistantsClient is deprecated. "
"Use OpenAIAssistantsClient (also deprecated) or migrate to OpenAIChatClient."
)
class AzureOpenAIAssistantsClient(
OpenAIAssistantsClient[AzureOpenAIAssistantsOptionsT], Generic[AzureOpenAIAssistantsOptionsT]
):
"""Deprecated Azure OpenAI Assistants client. Use OpenAIAssistantsClient or migrate to OpenAIChatClient."""
DEFAULT_AZURE_API_VERSION: ClassVar[str] = "2024-05-01-preview"
def __init__(
self,
*,
deployment_name: str | None = None,
assistant_id: str | None = None,
assistant_name: str | None = None,
assistant_description: str | None = None,
thread_id: str | None = None,
api_key: str | None = None,
endpoint: str | None = None,
base_url: str | None = None,
api_version: str | None = None,
token_endpoint: str | None = None,
credential: AzureCredentialTypes | AzureTokenProvider | None = None,
default_headers: Mapping[str, str] | None = None,
async_client: AsyncAzureOpenAI | None = None,
env_file_path: str | None = None,
env_file_encoding: str | None = None,
) -> None:
"""Initialize an Azure OpenAI Assistants client.
Keyword Args:
deployment_name: The Azure OpenAI deployment name.
assistant_id: The ID of an Azure OpenAI assistant to use.
assistant_name: The name to use when creating new assistants.
assistant_description: The description to use when creating new assistants.
thread_id: Default thread ID to use for conversations.
api_key: The API key to use.
endpoint: The deployment endpoint.
base_url: The deployment base URL.
api_version: The deployment API version.
token_endpoint: The token endpoint to request an Azure token.
credential: Azure credential or token provider for authentication.
default_headers: Default headers for HTTP requests.
async_client: An existing client to use.
env_file_path: Path to .env file for settings.
env_file_encoding: Encoding for .env file.
"""
azure_openai_settings = load_settings(
AzureOpenAISettings,
env_prefix="AZURE_OPENAI_",
api_key=api_key,
base_url=base_url,
endpoint=endpoint,
chat_deployment_name=deployment_name,
api_version=api_version,
env_file_path=env_file_path,
env_file_encoding=env_file_encoding,
token_endpoint=token_endpoint,
)
_apply_azure_defaults(azure_openai_settings, default_api_version=self.DEFAULT_AZURE_API_VERSION)
chat_deployment_name = azure_openai_settings.get("chat_deployment_name")
if not chat_deployment_name:
raise ValueError(
"Azure OpenAI deployment name is required. Set via 'deployment_name' parameter "
"or 'AZURE_OPENAI_CHAT_DEPLOYMENT_NAME' environment variable."
)
api_key_secret = azure_openai_settings.get("api_key")
token_scope = azure_openai_settings.get("token_endpoint")
ad_token_provider = None
if not async_client and not api_key_secret and credential:
ad_token_provider = resolve_credential_to_token_provider(credential, token_scope)
if not async_client and not api_key_secret and not ad_token_provider:
raise ValueError("Please provide either api_key, credential, or a client.")
if not async_client:
client_params: dict[str, Any] = {
"default_headers": default_headers,
}
if resolved_api_version := azure_openai_settings.get("api_version"):
client_params["api_version"] = resolved_api_version
if api_key_secret:
client_params["api_key"] = api_key_secret.get_secret_value()
elif ad_token_provider:
client_params["azure_ad_token_provider"] = ad_token_provider
if resolved_base_url := azure_openai_settings.get("base_url"):
client_params["base_url"] = str(resolved_base_url)
elif resolved_endpoint := azure_openai_settings.get("endpoint"):
client_params["azure_endpoint"] = str(resolved_endpoint)
async_client = AsyncAzureOpenAI(**client_params)
super().__init__(
model_id=chat_deployment_name,
assistant_id=assistant_id,
assistant_name=assistant_name,
assistant_description=assistant_description,
thread_id=thread_id,
async_client=async_client, # type: ignore[reportArgumentType]
default_headers=default_headers,
)
# endregion
# region AzureOpenAIEmbeddingClient
AzureOpenAIEmbeddingOptionsT = TypeVar(
"AzureOpenAIEmbeddingOptionsT",
bound=TypedDict, # type: ignore[valid-type]
default="OpenAIEmbeddingOptions",
covariant=True,
)
@deprecated("AzureOpenAIEmbeddingClient is deprecated. Use OpenAIEmbeddingClient with an AsyncAzureOpenAI client.")
class AzureOpenAIEmbeddingClient(
EmbeddingTelemetryLayer[str, list[float], AzureOpenAIEmbeddingOptionsT],
RawOpenAIEmbeddingClient[AzureOpenAIEmbeddingOptionsT],
Generic[AzureOpenAIEmbeddingOptionsT],
):
"""Deprecated Azure OpenAI embedding client. Use OpenAIEmbeddingClient with AsyncAzureOpenAI instead."""
OTEL_PROVIDER_NAME: ClassVar[str] = "azure.ai.openai"
def __init__(
self,
*,
api_key: str | None = None,
deployment_name: str | None = None,
endpoint: str | None = None,
base_url: str | None = None,
api_version: str | None = None,
token_endpoint: str | None = None,
credential: AzureCredentialTypes | AzureTokenProvider | None = None,
default_headers: Mapping[str, str] | None = None,
async_client: AsyncAzureOpenAI | None = None,
otel_provider_name: str | None = None,
env_file_path: str | None = None,
env_file_encoding: str | None = None,
) -> None:
"""Initialize an Azure OpenAI embedding client.
Keyword Args:
api_key: The API key.
deployment_name: The deployment name.
endpoint: The deployment endpoint.
base_url: The deployment base URL.
api_version: The deployment API version.
token_endpoint: The token endpoint to request an Azure token.
credential: Azure credential or token provider for authentication.
default_headers: Default headers for HTTP requests.
async_client: An existing client to use.
otel_provider_name: Override the OpenTelemetry provider name.
env_file_path: Path to .env file for settings.
env_file_encoding: Encoding for .env file.
"""
azure_openai_settings = load_settings(
AzureOpenAISettings,
env_prefix="AZURE_OPENAI_",
api_key=api_key,
base_url=base_url,
endpoint=endpoint,
embedding_deployment_name=deployment_name,
api_version=api_version,
env_file_path=env_file_path,
env_file_encoding=env_file_encoding,
token_endpoint=token_endpoint,
)
_apply_azure_defaults(azure_openai_settings)
embedding_deployment_name = azure_openai_settings.get("embedding_deployment_name")
if not embedding_deployment_name:
raise ValueError(
"Azure OpenAI embedding deployment name is required. Set via 'deployment_name' parameter "
"or 'AZURE_OPENAI_EMBEDDING_DEPLOYMENT_NAME' environment variable."
)
if not async_client:
# Create the Azure OpenAI client directly
merged_headers = dict(copy(default_headers)) if default_headers else {}
if APP_INFO:
merged_headers.update(APP_INFO)
merged_headers = prepend_agent_framework_to_user_agent(merged_headers)
api_key_secret = azure_openai_settings.get("api_key")
ad_token_provider = None
if not api_key_secret and credential:
ad_token_provider = resolve_credential_to_token_provider(
credential, azure_openai_settings.get("token_endpoint")
)
if not api_key_secret and not ad_token_provider:
raise ValueError("Please provide either api_key, credential, or a client.")
endpoint_value = azure_openai_settings.get("endpoint")
base_url_value = azure_openai_settings.get("base_url")
if not endpoint_value and not base_url_value:
raise ValueError("Please provide an endpoint or a base_url")
client_args: dict[str, Any] = {"default_headers": merged_headers}
if resolved_api_version := azure_openai_settings.get("api_version"):
client_args["api_version"] = resolved_api_version
if ad_token_provider:
client_args["azure_ad_token_provider"] = ad_token_provider
if api_key_secret:
client_args["api_key"] = api_key_secret.get_secret_value()
if base_url_value:
client_args["base_url"] = str(base_url_value)
if endpoint_value and not base_url_value:
client_args["azure_endpoint"] = str(endpoint_value)
if embedding_deployment_name:
client_args["azure_deployment"] = embedding_deployment_name
async_client = AsyncAzureOpenAI(**client_args)
# Store Azure-specific attributes for serialization
self.endpoint = str(azure_openai_settings.get("endpoint") or "")
self.api_version = azure_openai_settings.get("api_version") or ""
self.deployment_name = embedding_deployment_name
super().__init__(
async_client=async_client,
model=embedding_deployment_name,
default_headers=default_headers,
)
if otel_provider_name is not None:
self.OTEL_PROVIDER_NAME = otel_provider_name # type: ignore[misc]
# endregion
@@ -0,0 +1,67 @@
# Copyright (c) Microsoft. All rights reserved.
from __future__ import annotations
import logging
from collections.abc import Awaitable, Callable
from typing import Union
from agent_framework.exceptions import ChatClientInvalidAuthException
from azure.core.credentials import TokenCredential
from azure.core.credentials_async import AsyncTokenCredential
logger: logging.Logger = logging.getLogger(__name__)
AzureTokenProvider = Callable[[], Union[str, Awaitable[str]]]
"""A callable that returns a bearer token string, either synchronously or asynchronously."""
AzureCredentialTypes = Union[TokenCredential, AsyncTokenCredential]
"""Union of Azure credential types.
Accepts:
- ``TokenCredential`` — synchronous Azure credential (e.g. ``DefaultAzureCredential()``)
- ``AsyncTokenCredential`` — asynchronous Azure credential (e.g. ``azure.identity.aio.DefaultAzureCredential()``)
"""
def resolve_credential_to_token_provider(
credential: AzureCredentialTypes | AzureTokenProvider,
token_endpoint: str | None,
) -> AzureTokenProvider:
"""Convert an Azure credential or token provider into an ``ad_token_provider`` callable.
If the credential is already a callable token provider, it is returned as-is
(``token_endpoint`` is not required in this case).
If it is a ``TokenCredential`` or ``AsyncTokenCredential``, it is wrapped using
``azure.identity.get_bearer_token_provider`` (sync or async variant) which
handles token caching and automatic refresh.
Args:
credential: An Azure credential or token provider callable.
token_endpoint: The token scope/endpoint
(e.g. ``"https://cognitiveservices.azure.com/.default"``).
Required when ``credential`` is a ``TokenCredential`` or ``AsyncTokenCredential``.
Returns:
A callable that returns a bearer token string (sync or async).
Raises:
ServiceInvalidAuthError: If the token endpoint is empty when needed for credential wrapping.
"""
# Already a token provider callable (not a credential object) — use directly
if callable(credential) and not isinstance(credential, (TokenCredential, AsyncTokenCredential)):
return credential
if not token_endpoint:
raise ChatClientInvalidAuthException(
"A token endpoint must be provided either in settings, as an environment variable, or as an argument."
)
if isinstance(credential, AsyncTokenCredential):
from azure.identity.aio import get_bearer_token_provider as get_async_bearer_token_provider
return get_async_bearer_token_provider(credential, token_endpoint)
from azure.identity import get_bearer_token_provider
return get_bearer_token_provider(credential, token_endpoint) # type: ignore[arg-type]
@@ -1,261 +0,0 @@
# Copyright (c) Microsoft. All rights reserved.
"""Foundry Memory Context Provider using BaseContextProvider.
This module provides ``FoundryMemoryProvider``, built on
:class:`BaseContextProvider`.
"""
from __future__ import annotations
import logging
import sys
from contextlib import AbstractAsyncContextManager
from typing import TYPE_CHECKING, Any, ClassVar
from agent_framework import AGENT_FRAMEWORK_USER_AGENT, Message
from agent_framework._sessions import AgentSession, BaseContextProvider, SessionContext
from agent_framework._settings import load_settings
from agent_framework.azure._entra_id_authentication import AzureCredentialTypes
from azure.ai.projects.aio import AIProjectClient
from openai.types.responses import ResponseInputItemParam
from ._shared import AzureAISettings
if sys.version_info >= (3, 11):
from typing import Self # pragma: no cover
else:
from typing_extensions import Self # pragma: no cover
if TYPE_CHECKING:
from agent_framework._agents import SupportsAgentRun
logger = logging.getLogger(__name__)
class FoundryMemoryProvider(BaseContextProvider):
"""Foundry Memory context provider using the new BaseContextProvider hooks pattern.
Integrates Azure AI Foundry Memory Store for persistent semantic memory,
searching and storing memories via the Azure AI Projects SDK.
Args:
source_id: Unique identifier for this provider instance.
project_client: Azure AI Project client for memory operations.
memory_store_name: The name of the memory store to use.
scope: The namespace that logically groups and isolates memories (e.g., user ID).
context_prompt: The prompt to prepend to retrieved memories.
update_delay: Timeout period before processing memory update in seconds.
Defaults to 300 (5 minutes). Set to 0 to immediately trigger updates.
"""
DEFAULT_SOURCE_ID: ClassVar[str] = "foundry_memory"
DEFAULT_CONTEXT_PROMPT = "## Memories\nConsider the following memories when answering user questions:"
def __init__(
self,
source_id: str = DEFAULT_SOURCE_ID,
*,
project_client: AIProjectClient | None = None,
project_endpoint: str | None = None,
credential: AzureCredentialTypes | None = None,
allow_preview: bool | None = None,
memory_store_name: str,
scope: str | None = None,
context_prompt: str | None = None,
update_delay: int = 300,
env_file_path: str | None = None,
env_file_encoding: str | None = None,
) -> None:
"""Initialize the Foundry Memory context provider.
Args:
source_id: Unique identifier for this provider instance.
project_client: Azure AI Project client for memory operations.
project_endpoint: Azure AI project endpoint URL. Used when project_client is not provided.
credential: Azure credential for authentication. Accepts a TokenCredential,
AsyncTokenCredential, or a callable token provider.
Required when project_client is not provided.
allow_preview: Enables preview opt-in on internally-created ``AIProjectClient``.
memory_store_name: The name of the memory store to use.
scope: The namespace that logically groups and isolates memories (e.g., user ID).
If None, `session_id` will be used.
context_prompt: The prompt to prepend to retrieved memories.
update_delay: Timeout period before processing memory update in seconds.
env_file_path: Path to environment file for loading settings.
env_file_encoding: Encoding of the environment file.
"""
super().__init__(source_id)
azure_ai_settings = load_settings(
AzureAISettings,
env_prefix="AZURE_AI_",
project_endpoint=project_endpoint,
env_file_path=env_file_path,
env_file_encoding=env_file_encoding,
)
if project_client is None:
resolved_endpoint = azure_ai_settings.get("project_endpoint")
if not resolved_endpoint:
raise ValueError(
"Azure AI project endpoint is required. Set via 'project_endpoint' parameter "
"or 'AZURE_AI_PROJECT_ENDPOINT' environment variable."
)
if not credential:
raise ValueError("Azure credential is required when project_client is not provided.")
project_client_kwargs: dict[str, Any] = {
"endpoint": resolved_endpoint,
"credential": credential, # type: ignore[arg-type]
"user_agent": AGENT_FRAMEWORK_USER_AGENT,
}
if allow_preview is not None:
project_client_kwargs["allow_preview"] = allow_preview
project_client = AIProjectClient(**project_client_kwargs)
if not memory_store_name:
raise ValueError("memory_store_name is required")
if not scope:
raise ValueError("scope is required")
self.project_client = project_client
self.memory_store_name = memory_store_name
self.scope = scope
self.context_prompt = context_prompt or self.DEFAULT_CONTEXT_PROMPT
self.update_delay = update_delay
async def __aenter__(self) -> Self:
"""Async context manager entry."""
if self.project_client and isinstance(self.project_client, AbstractAsyncContextManager):
await self.project_client.__aenter__()
return self
async def __aexit__(self, exc_type: type[BaseException] | None, exc_val: BaseException | None, exc_tb: Any) -> None:
"""Async context manager exit."""
if self.project_client and isinstance(self.project_client, AbstractAsyncContextManager):
await self.project_client.__aexit__(exc_type, exc_val, exc_tb)
# -- Hooks pattern ---------------------------------------------------------
async def before_run(
self,
*,
agent: SupportsAgentRun,
session: AgentSession,
context: SessionContext,
state: dict[str, Any],
) -> None:
"""Search Foundry Memory for relevant memories and add to the session context.
This method:
1. Retrieves static memories (user profile) on first call per session
2. Searches for contextual memories based on input messages
3. Combines and injects memories into the context
"""
# On first run, retrieve static memories (user profile memories)
if not state.get("initialized"):
try:
static_search_result = await self.project_client.beta.memory_stores.search_memories(
name=self.memory_store_name,
scope=self.scope or context.session_id, # type: ignore[arg-type]
)
static_memories = [{"content": memory.memory_item.content} for memory in static_search_result.memories]
state["static_memories"] = static_memories
except Exception as e:
# Log but don't fail - memory retrieval is non-critical
logger.warning(f"Failed to retrieve static memories: {e}")
state["static_memories"] = []
finally:
# Mark as initialized regardless of success to avoid repeated attempts
state["initialized"] = True
# Search for contextual memories based on input messages
# Check if there are any non-empty input messages
has_input = any(msg and msg.text and msg.text.strip() for msg in context.input_messages)
if not has_input:
return
# Convert input messages to memory search item format
items: list[ResponseInputItemParam] = [
{"type": "message", "role": "user", "content": msg.text}
for msg in context.input_messages
if msg and msg.text and msg.text.strip()
]
try:
search_result = await self.project_client.beta.memory_stores.search_memories(
name=self.memory_store_name,
scope=self.scope or context.session_id, # type: ignore[arg-type]
items=items,
previous_search_id=state.get("previous_search_id"),
)
# Extract search_id for next incremental search
if search_result.memories:
state["previous_search_id"] = search_result.search_id
# Combine static and contextual memories
contextual_memories = [{"content": memory.memory_item.content} for memory in search_result.memories]
all_memories = state.get("static_memories", []) + contextual_memories
# Inject memories into context
if all_memories:
line_separated_memories = "\n".join(
str(memory.get("content", "")) for memory in all_memories if memory.get("content")
)
if line_separated_memories:
context.extend_messages(
self.source_id,
[Message(role="user", text=f"{self.context_prompt}\n{line_separated_memories}")],
)
except Exception as e:
# Log but don't fail - memory retrieval is non-critical
logger.warning(f"Failed to search contextual memories: {e}")
async def after_run(
self,
*,
agent: SupportsAgentRun,
session: AgentSession,
context: SessionContext,
state: dict[str, Any],
) -> None:
"""Store request/response messages to Foundry Memory for future retrieval.
This method updates the memory store with conversation messages.
The update is debounced by the configured update_delay.
"""
messages_to_store: list[Message] = list(context.input_messages)
if context.response and context.response.messages:
messages_to_store.extend(context.response.messages)
# Filter and convert messages to memory update item format
items: list[ResponseInputItemParam] = []
for message in messages_to_store:
if message.role in {"user", "assistant", "system"} and message.text and message.text.strip():
if message.role == "user":
items.append({"role": "user", "type": "message", "content": message.text})
elif message.role == "assistant":
items.append({"role": "assistant", "type": "message", "content": message.text})
if not items:
return
try:
# Fire and forget - don't wait for the update to complete
update_poller = await self.project_client.beta.memory_stores.begin_update_memories(
name=self.memory_store_name,
scope=self.scope or context.session_id, # type: ignore[arg-type]
items=items,
previous_update_id=state.get("previous_update_id"),
update_delay=self.update_delay,
)
# Store the update_id for next incremental update
state["previous_update_id"] = update_poller.update_id
except Exception as e:
# Log but don't fail - memory storage is non-critical
logger.warning(f"Failed to update memories: {e}")
__all__ = ["FoundryMemoryProvider"]
@@ -18,7 +18,6 @@ from agent_framework import (
from agent_framework._mcp import MCPTool
from agent_framework._settings import load_settings
from agent_framework._tools import ToolTypes
from agent_framework.azure._entra_id_authentication import AzureCredentialTypes
from azure.ai.projects.aio import AIProjectClient
from azure.ai.projects.models import (
AgentVersionDetails,
@@ -29,13 +28,15 @@ from azure.ai.projects.models import (
FunctionTool as AzureFunctionTool,
)
from ._client import AzureAIClient, AzureAIProjectAgentOptions
from ._client import AzureAIClient, AzureAIProjectAgentOptions # pyright: ignore[reportDeprecated]
from ._entra_id_authentication import AzureCredentialTypes
from ._shared import AzureAISettings, create_text_format_config, from_azure_ai_tools, to_azure_ai_tools
if sys.version_info >= (3, 13):
from typing import TypeVar # type: ignore # pragma: no cover
from warnings import deprecated # type: ignore # pragma: no cover
else:
from typing_extensions import TypeVar # type: ignore # pragma: no cover
from typing_extensions import TypeVar, deprecated # type: ignore # pragma: no cover
if sys.version_info >= (3, 11):
from typing import Self, TypedDict # type: ignore # pragma: no cover
else:
@@ -55,11 +56,12 @@ OptionsCoT = TypeVar(
)
@deprecated("AzureAIProjectAgentProvider is deprecated. Use FoundryAgent instead.")
class AzureAIProjectAgentProvider(Generic[OptionsCoT]):
"""Provider for Azure AI Agent Service (Responses API).
"""Deprecated provider for Azure AI Agent Service (Responses API).
This provider allows you to create, retrieve, and manage Azure AI agents
using the AIProjectClient from the Azure AI Projects SDK.
This provider is deprecated. Use ``FoundryAgent`` instead to connect to
pre-configured agents in Foundry.
Examples:
Using with explicit AIProjectClient:
@@ -200,7 +202,7 @@ class AzureAIProjectAgentProvider(Generic[OptionsCoT]):
)
# Extract options from default_options if present
opts = dict(default_options) if default_options else {}
opts: dict[str, Any] = dict(default_options) if default_options else {}
response_format = opts.get("response_format")
rai_config = opts.get("rai_config")
reasoning = opts.get("reasoning")
@@ -384,7 +386,7 @@ class AzureAIProjectAgentProvider(Generic[OptionsCoT]):
if not isinstance(details.definition, PromptAgentDefinition):
raise ValueError("Agent definition must be PromptAgentDefinition to get a Agent.")
client = AzureAIClient(
client = AzureAIClient( # pyright: ignore[reportDeprecated]
project_client=self._project_client,
agent_name=details.name,
agent_version=details.version,