import streamlit as st
import os
from langchain_huggingface import HuggingFaceEndpoint, ChatHuggingFace
from langchain_core.messages import HumanMessage, AIMessage, SystemMessage

hf = os.getenv('Data_science')
os.environ['HUGGINGFACEHUB_API_TOKEN'] = hf
os.environ['HF_TOKEN'] = hf

# --- Config ---
st.set_page_config(page_title="SQL Mentor Chat", layout="centered")
st.title("🗃️ SQL Mentor Chat")

# --- Sidebar for selections ---
st.sidebar.title("Mentor Preferences")
exp_options = ['Beginner', 'Intermediate', 'Experienced']
exp = st.sidebar.selectbox("Select your experience level:", exp_options)

# --- Initialize Chat Model ---
sql_model_skeleton = HuggingFaceEndpoint(
    repo_id='marin-community/marin-8b-instruct',
    provider='together',
    temperature=0.7,
    max_new_tokens=110,
    task='conversational'
)

sql_mentor = ChatHuggingFace(
    llm=sql_model_skeleton,
    repo_id='marin-community/marin-8b-instruct',
    provider='together',
    temperature=0.7,
    max_new_tokens=110,
    task='conversational'
)

# --- Session State ---
if "chat_history_sql" not in st.session_state:
    st.session_state.chat_history_sql = []

# --- Chat Form ---
with st.form(key="chat_form"):
    user_input = st.text_input("Ask your question:")
    submit = st.form_submit_button("Send")

# --- Chat Logic ---
if submit and user_input:
    system_prompt = (
        f"Act as a SQL mentor with {exp.lower()} experience level. "
        f"Explain in a very friendly manner and keep answers within 150 words. "
        f"If the question is not about SQL, politely say it is out of scope."
    )
    messages = [SystemMessage(content=system_prompt), HumanMessage(content=user_input)]
    result = sql_mentor.invoke(messages)
    st.session_state.chat_history_sql.append((user_input, result.content))

# --- Display Chat History ---
st.subheader("🗨️ Chat History")
for user, bot in st.session_state.chat_history_sql:
    st.markdown(f"**You:** {user}")
    st.markdown(f"**Mentor:** {bot}")
    st.markdown("---")