Create llm_handler.py

llm_handler.py

@@ -0,0 +1,85 @@
+# rentbot/llm_handler.py
+import os
+from openai import AsyncOpenAI
+
+client = AsyncOpenAI(api_key=os.getenv("OPENAI_API_KEY"))
+
+# Definition of the tool the LLM can use
+tools = [
+    {
+        "type": "function",
+        "function": {
+            "name": "create_event",
+            "description": "Create a calendar event to book an apartment viewing.",
+            "parameters": {
+                "type": "object",
+                "properties": {
+                    "start_time": {
+                        "type": "string",
+                        "description": "The start time of the event in ISO 8601 format, e.g., 2025-07-18T14:00:00",
+                    },
+                    "duration_minutes": {
+                        "type": "integer",
+                        "description": "The duration of the event in minutes.",
+                        "default": 30
+                    },
+                    "summary": {
+                        "type": "string",
+                        "description": "A short summary or name for the event, e.g., 'Unit 5B viewing'",
+                    },
+                },
+                "required": ["start_time", "summary"],
+            },
+        },
+    }
+]
+
+async def get_llm_response(messages: list):
+    """
+    Yields response chunks from OpenAI's chat completion streaming API.
+    Returns the full assistant message and any tool calls.
+    """
+    try:
+        stream = await client.chat.completions.create(
+            model="gpt-4o-mini",
+            messages=messages,
+            stream=True,
+            tools=tools,
+            tool_choice="auto",
+        )
+        
+        full_response = ""
+        tool_calls = []
+        
+        async for chunk in stream:
+            delta = chunk.choices[0].delta
+            if delta.content:
+                text_chunk = delta.content
+                full_response += text_chunk
+                yield text_chunk # Yield text chunks for real-time TTS
+            
+            if delta.tool_calls:
+                # Accumulate tool call chunks
+                if not tool_calls:
+                    tool_calls.extend(delta.tool_calls)
+                else:
+                    for i, tool_call_chunk in enumerate(delta.tool_calls):
+                        if tool_call_chunk.function.arguments:
+                            tool_calls[i].function.arguments += tool_call_chunk.function.arguments
+
+        # Construct the final assistant message object
+        assistant_message = {"role": "assistant", "content": full_response}
+        if tool_calls:
+            assistant_message["tool_calls"] = [
+                {
+                    "id": tc.id,
+                    "type": "function",
+                    "function": {"name": tc.function.name, "arguments": tc.function.arguments}
+                } for tc in tool_calls
+            ]
+            
+        return assistant_message, tool_calls
+
+    except Exception as e:
+        print(f"Error in get_llm_response: {e}")
+        return {"role": "assistant", "content": "I'm having a little trouble right now. Please try again in a moment."}, []