Re-use httpx client from assistants (#575)

To leverage connection pooling and reduce the number of connections from
assistant services to the workbench service

Author: markwaddle

assistants/project-assistant/assistant/conversation_clients.py

@@ -8,14 +8,12 @@
 
 from typing import Any, Optional, Tuple
 
-import semantic_workbench_api_model.workbench_service_client as wsc
 from semantic_workbench_api_model.workbench_service_client import ConversationAPIClient
-from semantic_workbench_assistant import settings
 from semantic_workbench_assistant.assistant_app import ConversationContext
 from semantic_workbench_assistant.storage import read_model
 
-from .logging import logger
 from .conversation_project_link import ConversationProjectManager
+from .logging import logger
 from .project_storage import ProjectStorageManager
 from .project_storage_models import ConversationRole
 
@@ -33,17 +31,7 @@ def get_conversation_client(context: ConversationContext, conversation_id: str)
         """
         Gets a client for accessing another conversation.
         """
-        # Create a client for the target conversation
-        builder = wsc.WorkbenchServiceClientBuilder(
-            base_url=str(settings.workbench_service_url),
-            assistant_service_id=context.assistant._assistant_service_id,
-            api_key=settings.workbench_service_api_key,
-        )
-        client = builder.for_conversation(
-            assistant_id=context.assistant.id,
-            conversation_id=conversation_id,
-        )
-        return client
+        return context.for_conversation(conversation_id)._conversation_client
 
     @staticmethod
     async def get_coordinator_client_for_project(
@@ -88,17 +76,10 @@ async def create_temporary_context_for_conversation(
         Creates a temporary context for the target conversation ID.
         """
         try:
-            # We'll use the same assistant as in the source context
-            assistant = source_context.assistant
-
             # Create a temporary context with the same properties as the original
             # but pointing to a different conversation
 
-            temp_context = ConversationContext(
-                assistant=assistant,
-                id=target_conversation_id,
-                title=source_context.title,
-            )
+            temp_context = source_context.for_conversation(target_conversation_id)
 
             return temp_context
 

examples/python/python-01-echo-bot/pyproject.toml

@@ -4,7 +4,7 @@ version = "0.1.0"
 description = "Example of a python Semantic Workbench assistant."
 authors = [{ name = "Semantic Workbench Team" }]
 readme = "README.md"
-requires-python = ">=3.11"
+requires-python = ">=3.11,<3.13"
 dependencies = ["openai>=1.61.0", "semantic-workbench-assistant>=0.1.0"]
 
 [tool.uv]

examples/python/python-01-echo-bot/uv.lock

@@ -4,7 +4,7 @@ version = "0.1.0"
 description = "Exploration of a python Semantic Workbench OpenAI assistant."
 authors = [{ name = "Semantic Workbench Team" }]
 readme = "README.md"
-requires-python = ">=3.11"
+requires-python = ">=3.11,<3.13"
 dependencies = [
     "openai>=1.61.0",
     "tiktoken>=0.8.0",

examples/python/python-02-simple-chatbot/pyproject.toml

@@ -4,7 +4,7 @@ version = "0.1.0"
 description = "Exploration of a python Semantic Workbench assistant that supports multiple AI models."
 authors = [{ name = "Semantic Workbench Team" }]
 readme = "README.md"
-requires-python = ">=3.11"
+requires-python = ">=3.11,<3.13"
 dependencies = [
     "anthropic>=0.34.2",
     "google-genai>=1.2.0",

examples/python/python-02-simple-chatbot/uv.lock

@@ -7,6 +7,7 @@
 from typing import Any, AsyncGenerator, AsyncIterator, Callable, Iterable
 from unittest import mock
 
+import httpx
 import pytest
 from assistant_extensions.attachments import AttachmentsConfigModel, AttachmentsExtension
 from llm_client.model import (
@@ -104,20 +105,23 @@ async def test_get_completion_messages_for_attachments(
 ) -> None:
     mock_assistant_app = mock.MagicMock(spec=AssistantAppProtocol)
 
+    assistant_id = uuid.uuid4()
+
     mock_conversation_context = mock.MagicMock(
         spec=ConversationContext(
             id="conversation_id",
             title="conversation_title",
             assistant=AssistantContext(
-                id="assistant_id",
+                id=str(assistant_id),
                 name="assistant_name",
                 _assistant_service_id="assistant_id",
                 _template_id="",
             ),
+            httpx_client=httpx.AsyncClient(),
         )
     )
     mock_conversation_context.id = "conversation_id"
-    mock_conversation_context.assistant.id = "assistant_id"
+    mock_conversation_context.assistant.id = str(assistant_id)
 
     mock_conversation_context.list_files.return_value = FileList(
         files=[

examples/python/python-03-multimodel-chatbot/pyproject.toml

@@ -177,11 +177,7 @@ async def duplicate_conversation(
         conversation_id = response.conversation_ids[0]
 
         # create a new conversation context
-        workflow_context = ConversationContext(
-            id=str(conversation_id),
-            title=title,
-            assistant=context.assistant,
-        )
+        workflow_context = context.for_conversation(str(conversation_id))
 
         # send link to chat for the new conversation
         await context.send_messages(

examples/python/python-03-multimodel-chatbot/uv.lock

@@ -4,7 +4,7 @@ version = "0.1.0"
 description = "Content Safety for Semantic Workbench Assistants"
 authors = [{ name = "Semantic Workbench Team" }]
 readme = "README.md"
-requires-python = ">=3.11"
+requires-python = ">=3.11,<3.13"
 dependencies = [
     "azure-ai-contentsafety>=1.0.0",
     "azure-core[aio]>=1.30.0",