Update app.py

app.py

@@ -1,8 +1,24 @@
-# Initialize Llama model (assuming you have a pre-trained model from llama)
-llama_model = ll.LlamaModel('your_llama_model_path')  # Replace with your Llama model path
+import yfinance as yf
+import gradio as gr
+import chatgroq  # Import Chatgroq API client
+from langchain.chains import LLMChain
+from langchain.memory import ConversationBufferMemory
+from langchain.prompts import PromptTemplate
 
-# Initialize Chatgroq (Configure accordingly based on your API or setup)
-chatgroq_model = chatgroq.ChatGroqModel('your_chatgroq_model_id')  # Replace with your Chatgroq ID
+# Initialize memory for conversation
+memory = ConversationBufferMemory(memory_key="chat_history", return_messages=True)
+
+# Define the prompt template for stock analysis
+prompt = """
+You are a stock market analyst. The user is asking for stock analysis.
+Here are the details of the stock:
+- Name: {stock_name}
+- Price: {stock_price}
+- Sector: {sector}
+- P/E Ratio: {pe_ratio}
+
+Analyze the stock performance and provide a brief summary and advice to the user.
+"""
 
 # Function to fetch stock data from Yahoo Finance
 def get_stock_data(ticker):
@@ -19,41 +35,38 @@ def get_stock_data(ticker):
         'pe_ratio': stock_pe_ratio
     }
 
-# Llama-based reasoning function to respond as a stock analyst
-def generate_analysis(stock_data):
-    prompt = f"""
-    You are a stock market analyst. 
-    Here are the details of the stock:
-    Name: {stock_data['stock_name']}
-    Price: {stock_data['stock_price']}
-    Sector: {stock_data['sector']}
-    P/E Ratio: {stock_data['pe_ratio']}
+# Function to generate stock analysis based on the conversation context
+def stock_analysis(ticker, user_query=""):
+    # Get API key from environment (no need to pass manually)
+    api_key = os.getenv("CHATGROQ_API_KEY")  # Ensure Hugging Face secret is set
     
-    Analyze the stock performance and provide a brief summary and advice to the user.
-    """
-    response = llama_model.generate_response(prompt)
-    return response
+    if not api_key:
+        return "API key is required for Chatgroq."
 
-# Chatgroq-based reasoning to understand context and provide advice
-def generate_advice_based_on_context(stock_data, user_query):
-    context = f"The stock price of {stock_data['stock_name']} is {stock_data['stock_price']}."
-    advice_prompt = f"User wants to know about the stock performance. {context} {user_query}"
-    advice = chatgroq_model.get_advice(advice_prompt)
-    return advice
+    # Initialize Chatgroq model with the API key
+    chatgroq_model = chatgroq.ChatGroqModel(api_key)
 
-# Gradio Interface function
-def stock_analysis(ticker, user_query=""):
+    # Set the stock data in the prompt template
     stock_data = get_stock_data(ticker)
-    analysis = generate_analysis(stock_data)
+    prompt_input = {
+        'stock_name': stock_data['stock_name'],
+        'stock_price': stock_data['stock_price'],
+        'sector': stock_data['sector'],
+        'pe_ratio': stock_data['pe_ratio']
+    }
+
+    # Create LLM chain with Chatgroq model and prompt
+    chain = LLMChain(llm=chatgroq_model, prompt=PromptTemplate.from_template(prompt), memory=memory)
     
-    # Generate user-specific advice (if any query is provided)
-    if user_query:
-        advice = generate_advice_based_on_context(stock_data, user_query)
-        return stock_data, analysis, advice
+    # Get analysis from Chatgroq-based agent
+    analysis = chain.run(input=prompt_input)
     
+    # Generate advice if there is a user query
+    if user_query:
+        return stock_data, analysis + "\n" + "Advice: " + user_query
     return stock_data, analysis
 
-# Create the Gradio interface
+# Gradio interface
 iface = gr.Interface(
     fn=stock_analysis,
     inputs=[
@@ -63,11 +76,8 @@ iface = gr.Interface(
     outputs=[
         gr.JSON(label="Stock Data"),
         gr.Textbox(label="Analysis"),
-        gr.Textbox(label="Advice", optional=True)
-    ],
-    live=True
+    ]
 )
 
-# Run the Gradio app
 if __name__ == '__main__':
     iface.launch()