Update app.py

app.py

@@ -2,22 +2,27 @@ import os
 import gradio as gr
 from smolagents import CodeAgent, tool, LiteLLMModel
 from qdrant_client import QdrantClient
-from qdrant_client.models import PointStruct, VectorParams, Distance, Filter, FieldCondition, MatchValue
+from qdrant_client.models import Filter, FieldCondition, MatchValue, VectorParams, Distance
 from sentence_transformers import SentenceTransformer
 
-# --- 1. Setup Connections ---
-# Ensure these match your Hugging Face Secrets names exactly
+# --- 1. Connection & Database Setup ---
+# These pull from your 'Secrets' on Hugging Face
 client = QdrantClient(
     url=os.getenv("QDRANT_URL"), 
     api_key=os.getenv("QDRANT_API_KEY")
 )
+
 embed_model = SentenceTransformer("all-MiniLM-L6-v2")
 collection_name = "cricket_analysis"
-
-# Starter Match Configuration
-STARTER_URL = "https://www.espncricinfo.com/series/icc-men-s-t20-world-cup-2024-1411166/south-africa-vs-india-final-1415755/ball-by-ball-commentary"
 STARTER_ID = "T20_WC_Final_2024"
 
+# Ensure the collection exists in Qdrant
+if not client.collection_exists(collection_name):
+    client.create_collection(
+        collection_name=collection_name,
+        vectors_config=VectorParams(size=384, distance=Distance.COSINE)
+    )
+
 # --- 2. The Search Tool ---
 @tool
 def search_cricket_database(query: str, active_match_id: str) -> str:
@@ -43,36 +48,45 @@ def search_cricket_database(query: str, active_match_id: str) -> str:
     
     return "\n\n".join([f"[Over {r.payload.get('over_number')}] {r.payload['text']}" for r in results])
 
-# --- 3. The Agent Brain ---
-# We define our professional rules here
-analyst_instructions = """
-You are a Lead Cricket Strategy Analyst. Your tone is professional and data-driven.
-Always use the 'search_cricket_database' tool to retrieve facts about the match.
-"""
+# --- 3. Defining the Brain & Agent ---
+# Define the model FIRST to avoid NameError 🧠
+model = LiteLLMModel(
+    model_id="meta-llama/Llama-3-70B-Instruct",
+    api_key=os.getenv("HF_TOKEN_cric")
+)
 
-# We pass the instructions as the 'description' or part of the first 'run'
+# Initialize the agent with the model and tools 🤖
 agent = CodeAgent(
     tools=[search_cricket_database], 
-    model=model,
-    # Remove system_prompt=system_prompt
+    model=model
 )
 
-# --- 4. Gradio UI Logic ---
+# --- 4. UI Logic ---
 def chat_fn(message, history, active_match_id):
-    prompt = f"Using context from match {active_match_id}: {message}"
-    # The agent.run command handles the tool calling and Llama logic
-    response = agent.run(prompt)
+    # Professional instructions injected here to maintain the persona 🎙️
+    system_instructions = (
+        "You are a Lead Cricket Strategy Analyst. Your tone is professional and data-driven. "
+        "Use the 'search_cricket_database' tool to retrieve facts about the match. "
+    )
+    
+    full_prompt = f"{system_instructions}\n\nUser Question about match {active_match_id}: {message}"
+    
+    # Run the agent and update history
+    response = agent.run(full_prompt)
     history.append((message, response))
     return "", history
 
-# --- 5. Interface Layout ---
+# --- 5. Gradio Interface Layout ---
 with gr.Blocks(theme=gr.themes.Soft()) as demo:
     gr.Markdown("# 🏏 Elite Cricket Analytics Portal")
     
-    active_id = gr.Textbox(label="Active Match Focus", value=STARTER_ID, interactive=False)
+    with gr.Row():
+        active_id = gr.Textbox(label="Active Match Focus", value=STARTER_ID, interactive=False)
+    
     chatbot = gr.Chatbot(label="Analysis Feed")
-    msg = gr.Textbox(label="Query the Analyst")
+    msg = gr.Textbox(label="Query the Analyst (e.g., 'How did the powerplay go?')")
 
     msg.submit(chat_fn, [msg, chatbot, active_id], [msg, chatbot])
 
-demo.launch()
+if __name__ == "__main__":
+    demo.launch()