src/streamlit_app.py

import streamlit as st
from langchain_community.utilities import SQLDatabase
from langchain.chat_models import ChatOpenAI
from langchain.agents import create_sql_agent
from langchain_groq import ChatGroq
from langchain_community.agent_toolkits import SQLDatabaseToolkit
from dotenv import load_dotenv
import tempfile
import sqlite3
import pandas as pd
import os

load_dotenv()

def is_valid_sqlite(file_path):
    try:
        with sqlite3.connect(file_path) as conn:
            conn.execute("SELECT name FROM sqlite_master LIMIT 1;")
        return True
    except sqlite3.DatabaseError:
        return False

def text_to_sql(query: str, db_path: str, llm_provider: str, api_key: str, model_name: str):
    try:
        db = SQLDatabase.from_uri(f"sqlite:///{db_path}")

        if llm_provider == 'OPENAI':
            llm = ChatOpenAI(api_key=api_key, model=model_name)
        elif llm_provider == 'OPEN_ROUTER':
            llm = ChatOpenAI(api_key=api_key, base_url='https://openrouter.ai/api/v1', model=model_name)
        elif llm_provider == 'GROQ':
            llm = ChatGroq(api_key=api_key, model=model_name)
        else:
            return "Unsupported LLM provider selected."

        toolkit = SQLDatabaseToolkit(llm=llm, db=db)
        db_chain = create_sql_agent(llm=llm, toolkit=toolkit, verbose=True)
        return db_chain.run(query)

    except Exception as e:
        return f"Error: {str(e)}"

def show_tables_as_df(db_path):
    conn = sqlite3.connect(db_path)
    cursor = conn.cursor()
    cursor.execute("SELECT name FROM sqlite_master WHERE type='table';")
    tables = cursor.fetchall()

    if tables:
        for table_name in tables:
            table = table_name[0]
            st.subheader(f"Table: {table}")
            df = pd.read_sql_query(f"SELECT * FROM {table} LIMIT 10", conn)
            st.dataframe(df)
    else:
        st.write("No tables found in database.")

    conn.close()


# Streamlit UI
st.title('🗃️ Chat with YOUR SQLite Database')

st.write("Upload your `.db` file and interact using natural language queries powered by LLMs.")

uploaded_file = st.file_uploader("Upload your `.db` file", type=["db"], accept_multiple_files=False)

llm_provider = st.radio("Choose LLM Provider", options=['OPEN_ROUTER', 'GROQ', 'OPENAI'])
model_name = st.text_input("Enter the Model Name", value='nousresearch/deephermes-3-mistral-24b-preview:free')
api_key = st.text_input("Enter Your API Key", type="password")
query = st.text_area("Enter Your Query")

if uploaded_file:
    with tempfile.NamedTemporaryFile(delete=False, suffix=".db", dir="/tmp") as tmpfile:
        tmpfile.write(uploaded_file.read())
        tmp_db_path = tmpfile.name

    if not is_valid_sqlite(tmp_db_path):
        st.error("Uploaded file is not a valid SQLite database.")
    else:
        st.success(f"Valid database `{uploaded_file.name}` uploaded!")

        st.info("Displaying first 10 rows from each table:")
        show_tables_as_df(tmp_db_path)

        if st.button("RUN Query"):
            if not api_key or not model_name:
                st.error("Please provide API key and model name.")
            elif not query.strip():
                st.error("Please enter a query.")
            else:
                st.info(f"Running query on `{uploaded_file.name}`...")
                result = text_to_sql(query, tmp_db_path, llm_provider, api_key, model_name)
                st.success("Query Result:")
                st.write(result)
else:
    st.info("Please upload a SQLite `.db` file to begin.")