Migrate AgentRunner to Agent Workflow (Python)

helpers/python.ts

@@ -470,12 +470,6 @@ export const installPythonTemplate = async ({
       }
     }
 
-    // Copy engine code
-    await copy("**", enginePath, {
-      parents: true,
-      cwd: path.join(compPath, "engines", "python", engine),
-    });
-
     // Copy router code
     await copyRouterCode(root, tools ?? []);
   }

questions/datasources.ts

@@ -19,10 +19,12 @@ export const getDataSourceChoices = (
     });
   }
   if (selectedDataSource === undefined || selectedDataSource.length === 0) {
-    choices.push({
-      title: "No datasource",
-      value: "none",
-    });
+    if (framework !== "fastapi") {
+      choices.push({
+        title: "No datasource",
+        value: "none",
+      });
+    }
     choices.push({
       title:
         process.platform !== "linux"

templates/types/streaming/fastapi/app/api/routers/chat.py

@@ -3,15 +3,16 @@
 from fastapi import APIRouter, BackgroundTasks, HTTPException, Request, status
 from llama_index.core.llms import MessageRole
 
-from app.api.routers.events import EventCallbackHandler
+from app.api.callbacks.llamacloud import LlamaCloudFileDownload
+from app.api.callbacks.next_question import SuggestNextQuestions
+from app.api.callbacks.stream_handler import StreamHandler
 from app.api.routers.models import (
     ChatData,
     Message,
     Result,
     SourceNodes,
 )
-from app.api.routers.vercel_response import VercelStreamResponse
-from app.engine.engine import get_chat_engine
+from app.engine.engine import get_engine
 from app.engine.query_filter import generate_filters
 
 chat_router = r = APIRouter()
@@ -36,15 +37,19 @@ async def chat(
         logger.info(
             f"Creating chat engine with filters: {str(filters)}",
         )
-        event_handler = EventCallbackHandler()
-        chat_engine = get_chat_engine(
-            filters=filters, params=params, event_handlers=[event_handler]
-        )
-        response = chat_engine.astream_chat(last_message_content, messages)
-
-        return VercelStreamResponse(
-            request, event_handler, response, data, background_tasks
+        engine = get_engine(filters=filters, params=params)
+        handler = engine.run(
+            user_msg=last_message_content,
+            chat_history=messages,
+            stream=True,
         )
+        return StreamHandler.from_default(
+            handler=handler,
+            callbacks=[
+                LlamaCloudFileDownload.from_default(background_tasks),
+                SuggestNextQuestions.from_default(data),
+            ],
+        ).vercel_stream()
     except Exception as e:
         logger.exception("Error in chat engine", exc_info=True)
         raise HTTPException(
@@ -53,25 +58,27 @@ async def chat(
         ) from e
 
 
-# non-streaming endpoint - delete if not needed
-@r.post("/request")
-async def chat_request(
-    data: ChatData,
-) -> Result:
-    last_message_content = data.get_last_message_content()
-    messages = data.get_history_messages()
+# TODO: Update non-streaming endpoint
+# Would be better if we use same chat.py endpoint for both agent and multiagent templates
+# # non-streaming endpoint - delete if not needed
+# @r.post("/request")
+# async def chat_request(
+#     data: ChatData,
+# ) -> Result:
+#     last_message_content = data.get_last_message_content()
+#     messages = data.get_history_messages()
 
-    doc_ids = data.get_chat_document_ids()
-    filters = generate_filters(doc_ids)
-    params = data.data or {}
-    logger.info(
-        f"Creating chat engine with filters: {str(filters)}",
-    )
+#     doc_ids = data.get_chat_document_ids()
+#     filters = generate_filters(doc_ids)
+#     params = data.data or {}
+#     logger.info(
+#         f"Creating chat engine with filters: {str(filters)}",
+#     )
 
-    chat_engine = get_chat_engine(filters=filters, params=params)
+#     chat_engine = get_chat_engine(filters=filters, params=params)
 
-    response = await chat_engine.achat(last_message_content, messages)
-    return Result(
-        result=Message(role=MessageRole.ASSISTANT, content=response.response),
-        nodes=SourceNodes.from_source_nodes(response.source_nodes),
-    )
+#     response = await chat_engine.achat(last_message_content, messages)
+#     return Result(
+#         result=Message(role=MessageRole.ASSISTANT, content=response.response),
+#         nodes=SourceNodes.from_source_nodes(response.source_nodes),
+#     )

templates/types/streaming/fastapi/app/api/routers/events.py

@@ -99,6 +99,8 @@ def to_response(self):
             return None
 
 
+# TODO: Add an adapter for workflow events
+# and remove callback handler
 class EventCallbackHandler(BaseCallbackHandler):
     _aqueue: asyncio.Queue
     is_done: bool = False