mental_support.py

import streamlit as st
import ollama
import base64

# Set page config
st.set_page_config(page_title="Mental Health Chatbot")

# Function to encode image to base64
def get_base64(background):
    with open(background, "rb") as f:
        data = f.read()
    return base64.b64encode(data).decode()

# Encode the background image
bin_str = get_base64("background.jpg")
print("Base64 String:", bin_str)  # Debugging

# Apply the background image using CSS
st.markdown(f"""
    <style>
        .stApp {{
            background-image: url("data:img/png;base64,{bin_str}");
            background-size: cover;
            background-position: center;
            background-repeat: no-repeat;
        }}
    </style>
    """, unsafe_allow_html=True)

# Set the title color explicitly
st.markdown(
    "<h1 style='color: #432B2B; text-align: center;'>Mental Health Support Agent</h1>",
    unsafe_allow_html=True
)

# Initialize conversation history
st.session_state.setdefault('conversation_history', [])

def generate_response(user_input):
    # Add user input to conversation history
    st.session_state['conversation_history'].append({"role": "user", "content": user_input})
    
    # Generate AI response using Ollama
    response = ollama.chat(model="llama3:8b", messages=st.session_state['conversation_history'])
    print("Received response from Ollama:", response)  # Debugging
    ai_response = response['message']['content']
    
    # Add AI response to conversation history
    st.session_state['conversation_history'].append({"role": "assistant", "content": ai_response})
    return ai_response

def generate_affirmation():
    prompt = "Provide a positive affirmation to encourage someone who is feeling stressed or overwhelmed"    
    response = ollama.chat(model="llama3:8b", messages=[{"role": "user", "content": prompt}])
    return response['message']['content']

def generate_meditation_guide():
    prompt = "Provide a 5-minute guided meditation script to help someone relax and reduce stress."    
    response = ollama.chat(model="llama3:8b", messages=[{"role": "user", "content": prompt}])
    return response['message']['content']

# Display conversation history
for msg in st.session_state['conversation_history']:
    role = "You" if msg['role'] == "user" else "AI"
    st.markdown(f"**{role}:** {msg['content']}")

# User input
user_message = st.text_input("How can I help you today?")

if user_message:
    with st.spinner("Thinking....."):
        ai_response = generate_response(user_message)
        st.markdown(f"**AI:** {ai_response}")

col1, col2 = st.columns(2)

with col1:
    if st.button("Gift me a positive Affirmation"):
        affirmation = generate_affirmation()
        st.markdown(f"**Affirmation:** {affirmation}")

with col2:
    if st.button("Gift me a guided meditation"):
        meditation_guide = generate_meditation_guide()
        st.markdown(f"**Guided Meditation:** {meditation_guide}")