import os import pandas as pd import streamlit as st from pandasai import SmartDataframe, SmartDatalake from pandasai.llm import GooglePalm from langchain_google_genai import ChatGoogleGenerativeAI from langchain_experimental.agents import create_pandas_dataframe_agent # from pandasai.llm.openai import OpenAI from langchain_groq import ChatGroq def read_csv_file(uploaded_file): try: df = pd.read_csv(uploaded_file) except UnicodeDecodeError: st.warning("UTF-8 encoding failed. Trying latin1 encoding.") try: df = pd.read_csv(uploaded_file, encoding='latin1') print(uploaded_file) except UnicodeDecodeError: st.error("Unable to read the file with both UTF-8 and ISO-8859-1 encodings.") return None return df def main(): st.set_page_config(page_title="Smart Data Query App", layout="wide") st.title("Smart Data Query App") # Choice to upload one or two CSV files file_count = st.radio("How many CSV files would you like to upload?", (1, 2)) col1, col2 = st.columns(2) with col1: uploaded_file1 = st.file_uploader("Choose the first CSV file", type="csv", key="file1") if file_count == 2: with col2: uploaded_file2 = st.file_uploader("Choose the second CSV file", type="csv", key="file2") else: uploaded_file2 = None if uploaded_file1 is not None or uploaded_file2 is not None: df1 = read_csv_file(uploaded_file1) if uploaded_file1 is not None else None df2 = read_csv_file(uploaded_file2) if uploaded_file2 is not None else None if df1 is not None or df2 is not None: if df1 is not None: with st.expander("Preview First CSV File"): st.dataframe(df1.head()) if df2 is not None: with st.expander("Preview Second CSV File"): st.dataframe(df2.head()) pandas_api = os.environ['PANDASAI_API_KEY'] google_api = os.environ['GOOGLE_API_KEY'] groq_api = os.environ['GROQ_API_KEY'] # Set up the ChatGroq model llm = ChatGroq( groq_api_key=groq_api, model_name='mixtral-8x7b-32768' ) # llm = GooglePalm(api_key=google_api) if df1 is not None and df2 is not None: lake = SmartDatalake([df1, df2]) else: lake = SmartDataframe(df1, config = {"LLM": llm, "conversational": True, "verbose": True}) if df1 is not None else None datalake_1 = lake query = st.text_input("Enter your query:") submitted = st.button("Submit") if submitted: if query: response = datalake_1.chat(query) if "Unfortunately, I was not able to answer your question, because of the following error:" in response: llm = ChatGoogleGenerativeAI( model="gemini-pro", verbose=True, google_api_key=google_api ) agent = create_pandas_dataframe_agent( llm, df1, verbose=True, ) response2 = agent.invoke(query) st.write("Response:") st.write(response2["output"]) else: st.write("Response:") st.write(response) # Append the query and response to the session state for query history if 'query_history' not in st.session_state: st.session_state.query_history = [] st.session_state.query_history.append((query, response)) else: st.write("Please enter a query.") # Display query history if 'query_history' in st.session_state and st.session_state.query_history: st.subheader("Query History") for q, r in st.session_state.query_history: st.write(f"**Query:** {q}") st.write(f"**Response:** {r}") st.write("---") else: st.error("Failed to read one or both CSV files. Please check the files and try again.") else: st.write("Please upload at least one CSV file.") if __name__ == "__main__": main()