Merge branch 'main' into fix-studio-pyproject-toml

README.md

@@ -25,6 +25,13 @@ pip install -U "autogen-agentchat" "autogen-ext[openai]"
 
 The current stable version is v0.4. If you are upgrading from AutoGen v0.2, please refer to the [Migration Guide](https://microsoft.github.io/autogen/dev/user-guide/agentchat-user-guide/migration-guide.html) for detailed instructions on how to update your code and configurations.
 
+```bash
+# Install AutoGen Studio for no-code GUI
+pip install -U "autogenstudio"
+```
+
+## Quickstart
+
 ### Hello World
 
 Create an assistant agent using OpenAI's GPT-4o model.
@@ -69,6 +76,15 @@ async def main() -> None:
 asyncio.run(main())
 ```
 
+### AutoGen Studio
+
+Use AutoGen Studio to prototype and run multi-agent workflows without writing code.
+
+```bash
+# Run AutoGen Studio on http://localhost:8080
+autogenstudio ui --port 8080 --appdir ./my-app
+```
+
 ## Why Use AutoGen?
 
 <div align="center">

python/packages/autogen-agentchat/src/autogen_agentchat/agents/_user_proxy_agent.py

@@ -1,15 +1,17 @@
 import asyncio
+import uuid
+from contextlib import contextmanager
+from contextvars import ContextVar
 from inspect import iscoroutinefunction
-from typing import Awaitable, Callable, Optional, Sequence, Union, cast
+from typing import Any, AsyncGenerator, Awaitable, Callable, ClassVar, Generator, Optional, Sequence, Union, cast
 
 from aioconsole import ainput  # type: ignore
 from autogen_core import CancellationToken
 
 from ..base import Response
-from ..messages import ChatMessage, HandoffMessage, TextMessage
+from ..messages import AgentEvent, ChatMessage, HandoffMessage, TextMessage, UserInputRequestedEvent
 from ._base_chat_agent import BaseChatAgent
 
-# Define input function types more precisely
 SyncInputFunc = Callable[[str], str]
 AsyncInputFunc = Callable[[str, Optional[CancellationToken]], Awaitable[str]]
 InputFuncType = Union[SyncInputFunc, AsyncInputFunc]
@@ -109,6 +111,33 @@ async def cancellable_user_agent():
                     print(f"BaseException: {e}")
     """
 
+    class InputRequestContext:
+        def __init__(self) -> None:
+            raise RuntimeError(
+                "InputRequestContext cannot be instantiated. It is a static class that provides context management for user input requests."
+            )
+
+        _INPUT_REQUEST_CONTEXT_VAR: ClassVar[ContextVar[str]] = ContextVar("_INPUT_REQUEST_CONTEXT_VAR")
+
+        @classmethod
+        @contextmanager
+        def populate_context(cls, ctx: str) -> Generator[None, Any, None]:
+            """:meta private:"""
+            token = UserProxyAgent.InputRequestContext._INPUT_REQUEST_CONTEXT_VAR.set(ctx)
+            try:
+                yield
+            finally:
+                UserProxyAgent.InputRequestContext._INPUT_REQUEST_CONTEXT_VAR.reset(token)
+
+        @classmethod
+        def request_id(cls) -> str:
+            try:
+                return cls._INPUT_REQUEST_CONTEXT_VAR.get()
+            except LookupError as e:
+                raise RuntimeError(
+                    "InputRequestContext.runtime() must be called within the input callback of a UserProxyAgent."
+                ) from e
+
     def __init__(
         self,
         name: str,
@@ -153,9 +182,15 @@ async def _get_input(self, prompt: str, cancellation_token: Optional[Cancellatio
         except Exception as e:
             raise RuntimeError(f"Failed to get user input: {str(e)}") from e
 
-    async def on_messages(
-        self, messages: Sequence[ChatMessage], cancellation_token: Optional[CancellationToken] = None
-    ) -> Response:
+    async def on_messages(self, messages: Sequence[ChatMessage], cancellation_token: CancellationToken) -> Response:
+        async for message in self.on_messages_stream(messages, cancellation_token):
+            if isinstance(message, Response):
+                return message
+        raise AssertionError("The stream should have returned the final result.")
+
+    async def on_messages_stream(
+        self, messages: Sequence[ChatMessage], cancellation_token: CancellationToken
+    ) -> AsyncGenerator[AgentEvent | ChatMessage | Response, None]:
         """Handle incoming messages by requesting user input."""
         try:
             # Check for handoff first
@@ -164,15 +199,18 @@ async def on_messages(
                 f"Handoff received from {handoff.source}. Enter your response: " if handoff else "Enter your response: "
             )
 
-            user_input = await self._get_input(prompt, cancellation_token)
+            request_id = str(uuid.uuid4())
+
+            input_requested_event = UserInputRequestedEvent(request_id=request_id, source=self.name)
+            yield input_requested_event
+            with UserProxyAgent.InputRequestContext.populate_context(request_id):
+                user_input = await self._get_input(prompt, cancellation_token)
 
             # Return appropriate message type based on handoff presence
             if handoff:
-                return Response(
-                    chat_message=HandoffMessage(content=user_input, target=handoff.source, source=self.name)
-                )
+                yield Response(chat_message=HandoffMessage(content=user_input, target=handoff.source, source=self.name))
             else:
-                return Response(chat_message=TextMessage(content=user_input, source=self.name))
+                yield Response(chat_message=TextMessage(content=user_input, source=self.name))
 
         except asyncio.CancelledError:
             raise

python/packages/autogen-agentchat/src/autogen_agentchat/messages.py

@@ -103,25 +103,40 @@ class ToolCallSummaryMessage(BaseChatMessage):
     type: Literal["ToolCallSummaryMessage"] = "ToolCallSummaryMessage"
 
 
+class UserInputRequestedEvent(BaseAgentEvent):
+    """An event signaling a that the user proxy has requested user input. Published prior to invoking the input callback."""
+
+    request_id: str
+    """Identifier for the user input request."""
+
+    content: Literal[""] = ""
+    """Empty content for compat with consumers expecting a content field."""
+
+    type: Literal["UserInputRequestedEvent"] = "UserInputRequestedEvent"
+
+
 ChatMessage = Annotated[
     TextMessage | MultiModalMessage | StopMessage | ToolCallSummaryMessage | HandoffMessage, Field(discriminator="type")
 ]
 """Messages for agent-to-agent communication only."""
 
 
-AgentEvent = Annotated[ToolCallRequestEvent | ToolCallExecutionEvent, Field(discriminator="type")]
+AgentEvent = Annotated[
+    ToolCallRequestEvent | ToolCallExecutionEvent | UserInputRequestedEvent, Field(discriminator="type")
+]
 """Events emitted by agents and teams when they work, not used for agent-to-agent communication."""
 
 
 __all__ = [
+    "AgentEvent",
     "BaseMessage",
-    "TextMessage",
+    "ChatMessage",
+    "HandoffMessage",
     "MultiModalMessage",
     "StopMessage",
-    "HandoffMessage",
-    "ToolCallRequestEvent",
+    "TextMessage",
     "ToolCallExecutionEvent",
+    "ToolCallRequestEvent",
     "ToolCallSummaryMessage",
-    "ChatMessage",
-    "AgentEvent",
+    "UserInputRequestedEvent",
 ]

python/packages/autogen-agentchat/src/autogen_agentchat/ui/__init__.py

@@ -2,6 +2,6 @@
 This module implements utility classes for formatting/printing agent messages.
 """
 
-from ._console import Console
+from ._console import Console, UserInputManager
 
-__all__ = ["Console"]
+__all__ = ["Console", "UserInputManager"]

python/packages/autogen-agentchat/src/autogen_agentchat/ui/_console.py

@@ -1,14 +1,17 @@
+import asyncio
 import os
 import sys
 import time
-from typing import AsyncGenerator, List, Optional, TypeVar, cast
+from inspect import iscoroutinefunction
+from typing import AsyncGenerator, Awaitable, Callable, Dict, List, Optional, TypeVar, Union, cast
 
 from aioconsole import aprint  # type: ignore
-from autogen_core import Image
+from autogen_core import CancellationToken, Image
 from autogen_core.models import RequestUsage
 
+from autogen_agentchat.agents import UserProxyAgent
 from autogen_agentchat.base import Response, TaskResult
-from autogen_agentchat.messages import AgentEvent, ChatMessage, MultiModalMessage
+from autogen_agentchat.messages import AgentEvent, ChatMessage, MultiModalMessage, UserInputRequestedEvent
 
 
 def _is_running_in_iterm() -> bool:
@@ -19,14 +22,60 @@ def _is_output_a_tty() -> bool:
     return sys.stdout.isatty()
 
 
+SyncInputFunc = Callable[[str], str]
+AsyncInputFunc = Callable[[str, Optional[CancellationToken]], Awaitable[str]]
+InputFuncType = Union[SyncInputFunc, AsyncInputFunc]
+
 T = TypeVar("T", bound=TaskResult | Response)
 
 
+class UserInputManager:
+    def __init__(self, callback: InputFuncType):
+        self.input_events: Dict[str, asyncio.Event] = {}
+        self.callback = callback
+
+    def get_wrapped_callback(self) -> AsyncInputFunc:
+        async def user_input_func_wrapper(prompt: str, cancellation_token: Optional[CancellationToken]) -> str:
+            # Lookup the event for the prompt, if it exists wait for it.
+            # If it doesn't exist, create it and store it.
+            # Get request ID:
+            request_id = UserProxyAgent.InputRequestContext.request_id()
+            if request_id in self.input_events:
+                event = self.input_events[request_id]
+            else:
+                event = asyncio.Event()
+                self.input_events[request_id] = event
+
+            await event.wait()
+
+            del self.input_events[request_id]
+
+            if iscoroutinefunction(self.callback):
+                # Cast to AsyncInputFunc for proper typing
+                async_func = cast(AsyncInputFunc, self.callback)
+                return await async_func(prompt, cancellation_token)
+            else:
+                # Cast to SyncInputFunc for proper typing
+                sync_func = cast(SyncInputFunc, self.callback)
+                loop = asyncio.get_event_loop()
+                return await loop.run_in_executor(None, sync_func, prompt)
+
+        return user_input_func_wrapper
+
+    def notify_event_received(self, request_id: str) -> None:
+        if request_id in self.input_events:
+            self.input_events[request_id].set()
+        else:
+            event = asyncio.Event()
+            self.input_events[request_id] = event
+
+
 async def Console(
     stream: AsyncGenerator[AgentEvent | ChatMessage | T, None],
     *,
     no_inline_images: bool = False,
     output_stats: bool = False,
+    user_input_manager: UserInputManager | None = None,
 ) -> T:
     """
     Consumes the message stream from :meth:`~autogen_agentchat.base.TaskRunner.run_stream`
@@ -67,6 +116,7 @@ async def Console(
                     f"Duration: {duration:.2f} seconds\n"
                 )
                 await aprint(output, end="")
+
             # mypy ignore
             last_processed = message  # type: ignore
 
@@ -96,9 +146,13 @@ async def Console(
                     f"Duration: {duration:.2f} seconds\n"
                 )
                 await aprint(output, end="")
+
             # mypy ignore
             last_processed = message  # type: ignore
-
+        # We don't want to print UserInputRequestedEvent messages, we just use them to signal the user input event.
+        elif isinstance(message, UserInputRequestedEvent):
+            if user_input_manager is not None:
+                user_input_manager.notify_event_received(message.request_id)
         else:
             # Cast required for mypy to be happy
             message = cast(AgentEvent | ChatMessage, message)  # type: ignore

python/packages/autogen-core/docs/src/index.md

@@ -84,7 +84,7 @@ Built on AgentChat.
 
 ```bash
 pip install autogenstudio
-autogenstudio ui --port 8080
+autogenstudio ui --port 8080 --appdir ./myapp
 ```
 
 +++

python/packages/autogen-core/docs/src/user-guide/agentchat-user-guide/tutorial/teams.ipynb

@@ -83,7 +83,7 @@
             "source": [
                 "## Running a Team\n",
                 "\n",
-                "Let's calls the {py:meth}`~autogen_agentchat.teams.BaseGroupChat.run` method\n",
+                "Let's call the {py:meth}`~autogen_agentchat.teams.BaseGroupChat.run` method\n",
                 "to start the team with a task."
             ]
         },

python/packages/autogen-core/src/autogen_core/models/_types.py

@@ -52,7 +52,7 @@ class RequestUsage:
     completion_tokens: int
 
 
-FinishReasons = Literal["stop", "length", "function_calls", "content_filter"]
+FinishReasons = Literal["stop", "length", "function_calls", "content_filter", "unknown"]
 
 
 @dataclass

python/packages/autogen-ext/src/autogen_ext/auth/azure/__init__.py

@@ -16,7 +16,7 @@ class TokenProviderConfig(BaseModel):
 class AzureTokenProvider(Component[TokenProviderConfig]):
     component_type = "token_provider"
     component_config_schema = TokenProviderConfig
-    component_provider_override = "autogen_ext.models.openai.AzureTokenProvider"
+    component_provider_override = "autogen_ext.auth.azure.AzureTokenProvider"
 
     def __init__(self, credential: TokenProvider, *scopes: str):
         self.credential = credential

python/packages/autogen-ext/src/autogen_ext/models/openai/_openai_client.py

@@ -30,6 +30,7 @@
     Image,
     MessageHandlerContext,
 )
+from autogen_core.models import FinishReasons
 from autogen_core.logging import LLMCallEvent
 from autogen_core.models import (
     AssistantMessage,
@@ -327,6 +328,21 @@ def assert_valid_name(name: str) -> str:
     return name
 
 
+def normalize_stop_reason(stop_reason: str | None) -> FinishReasons:
+    if stop_reason is None:
+        return "unknown"
+
+    # Convert to lower case
+    stop_reason = stop_reason.lower()
+
+    KNOWN_STOP_MAPPINGS: Dict[str, FinishReasons] = {
+        "end_turn": "stop",
+        "tool_calls": "function_calls",
+    }
+
+    return KNOWN_STOP_MAPPINGS.get(stop_reason, "unknown")
+
+
 class BaseOpenAIChatCompletionClient(ChatCompletionClient):
     def __init__(
         self,
@@ -747,8 +763,8 @@ async def create_stream(
         else:
             prompt_tokens = 0
 
-        if stop_reason is None:
-            raise ValueError("No stop reason found")
+        if stop_reason == "function_call":
+            raise ValueError("Function calls are not supported in this context")
 
         content: Union[str, List[FunctionCall]]
         if len(content_deltas) > 1:
@@ -770,13 +786,9 @@ async def create_stream(
             prompt_tokens=prompt_tokens,
             completion_tokens=completion_tokens,
         )
-        if stop_reason == "function_call":
-            raise ValueError("Function calls are not supported in this context")
-        if stop_reason == "tool_calls":
-            stop_reason = "function_calls"
 
         result = CreateResult(
-            finish_reason=stop_reason,  # type: ignore
+            finish_reason=normalize_stop_reason(stop_reason),
             content=content,
             usage=usage,
             cached=False,
@@ -1102,7 +1114,7 @@ class AzureOpenAIChatCompletionClient(
                 "azure_deployment": "{your-azure-deployment}",
                 "api_version": "2024-06-01",
                 "azure_ad_token_provider": {
-                    "provider": "autogen_ext.models.openai.AzureTokenProvider",
+                    "provider": "autogen_ext.auth.azure.AzureTokenProvider",
                     "config": {
                         "provider_kind": "DefaultAzureCredential",
                         "scopes": ["https://cognitiveservices.azure.com/.default"],