fix: chat agent integration, llama inference bug

backend/src/agent.py

@@ -2,6 +2,7 @@ import re
 import json
 import inspect
 import time
+import asyncio
 
 from temporary import generate_tmp_id
 from stream import Stream
@@ -79,12 +80,13 @@ class Agent:
 
         await self.run(run)
 
-        async def on_output_event(data):
+        async def on_output_event(stream):
+            print("[agent] run output " + repr(stream))
             new_message = await self.app.send_chat_message(
                 sender=self.id,
                 chat_id=message.chat_id,
                 thread_id=message.thread_id,
-                content=data.stream,
+                content=stream,
             )
             run.emit_event('chat-message', {
                 'messageId': new_message['id'],
@@ -92,7 +94,7 @@ class Agent:
                 'threadId': message.thread_id,
             })
 
-        run.on('output', on_output_event)
+        await run.on_async('output', on_output_event)
         return run
 
     def _role_for_message(self, message):
@@ -180,6 +182,12 @@ class AgentRun:
         for event in self.events:
             if event["name"] == event_name:
                 callback(event["data"])
+    
+    async def on_async(self, event_name, callback):
+        event_loop = asyncio.get_event_loop()
+        def new_callback(data):
+            event_loop.create_task(callback(data))
+        self.on(event_name, new_callback)
 
     def emit_event(self, event_name, data=None):
         """Record an event and notify existing listeners.

backend/src/inference.py

@@ -71,7 +71,7 @@ def build_messages(system_prompt="", instruction=None, input=None, messages=None
             raise Exception("Unsupported message type: " + repr(type(message)))
 
     # shouldn't be necessary. maybe an assert instead?
-    messages = [normalize(msg) for msg in messages]
+    messages = [normalize(msg) for msg in messages or []]
     # default user message
     prompt = input
 
@@ -141,7 +141,7 @@ async def chat_completion(app,
 # llama.cpp
 async def local_chat_completion(app, messages, model, tools=[], opts={}, agent_run=None):
     if agent_run:
-        content_stream = agent_run.stream
+        content_stream = agent_run.get_output_stream()
     else:
         content_stream = Stream(app)