Update app.py

app.py

@@ -7,6 +7,60 @@ import streamlit as st
 
 VECTOR_DB ="c8af7dfa-bcad-46e5-b69d-cd85ce9315d1"
 
+IBM_API_KEY = os.getenv("IBM_API_KEY")
+IBM_PROJECT_ID = "a0659778-f4ce-4da1-ba01-43b4f43a026f"
+
+IBM_URL_TOKEN = "https://iam.cloud.ibm.com/identity/token"
+IBM_URL_CHAT = "https://us-south.ml.cloud.ibm.com/ml/v1/text/chat?version=2023-10-25"
+
+##############################################
+##
+##   IBM API
+##
+##############################################
+def IBM_token():
+    # Define the headers
+    headers = {
+        "Content-Type": "application/x-www-form-urlencoded"
+    }
+    
+    # Define the data payload
+    data = {
+        "grant_type": "urn:ibm:params:oauth:grant-type:apikey",
+        "apikey": IBM_API_KEY
+    }
+    
+    # Make the POST request
+    response = requests.post(IBM_URL_TOKEN, headers=headers, data=data)
+    st.session_state.IBM_ACCESS_TOKEN = response.json().get("access_token", "")
+
+
+def IBM_chat (messages):
+    body = {
+        "model_id": "ibm/granite-3-8b-instruct",
+        "project_id": IBM_PROJECT_ID,
+        "messages": messages,
+        "max_tokens": 10000,
+        "temperature": 0.7,
+        "time_limit": 50000
+    }
+    headers = {
+    	"Accept": "application/json",
+    	"Content-Type": "application/json",
+    	"Authorization": "Bearer " + st.session_state.IBM_ACCESS_TOKEN
+    }    
+    response = requests.post(
+    	IBM_URL_CHAT,
+    	headers=headers,
+    	json=body
+    )
+    
+    if response.status_code != 200:
+    	raise Exception("Non-200 response: " + str(response.text))
+    
+    response = response.json()
+    return response["choices"][0]["message"]["content"]
+
 def get_credentials():
 	return {
 		"url" : "https://us-south.ml.cloud.ibm.com",
@@ -150,28 +204,25 @@ def proximity_search( question ):
 st.title("🔍 IBM Watson RAG Chatbot")
 
 # User input in Streamlit
-question = st.text_input("Enter your question:")
+user_input = st.text_input("Enter your question:")
 
-if question:
-    # Retrieve relevant grounding context
-    grounding = proximity_search(question)
+if user_input:
 
-    # Format the question with retrieved context
-    formatted_question = f"""<|start_of_role|>user<|end_of_role|>Use the following pieces of context to answer the question.
-{grounding}
-Question: {question}<|end_of_text|>
-<|start_of_role|>assistant<|end_of_role|>"""
+    # Display user message
+    st.chat_message("user").markdown(user_input)
 
-    # Placeholder for a prompt input (Optional)
-    prompt_input = ""  # Set this dynamically if needed
-    prompt = f"""{prompt_input}{formatted_question}"""
+    grounding = proximity_search(user_input)
 
-    # Simulated AI response (Replace with actual model call)
-    generated_response = f"AI Response based on: {prompt}"
+    # add the submissions as context (only in prompt, not in history)
+    prompt = user_input + ". Provide urls where possible. Given the context: " + grounding;
+    messages = st.session_state.messages.copy()
+    messages.append({"role": "user", "content": prompt})
+    st.session_state.messages.append({"role": "user", "content": user_input})
 
-    # Display results
-    st.subheader("📌 Retrieved Context")
-    st.write(grounding)
+    # Get response from IBM
+    with st.spinner("Thinking..."):
+        assistant_reply = IBM_chat(messages)
 
-    st.subheader("🤖 AI Response")
-    st.write(generated_response)
+    # Display assistant message
+    st.chat_message("assistant").markdown(assistant_reply)
+    st.session_state.messages.append({"role": "assistant", "content": assistant_reply})