app.py

import streamlit as st
import pandas as pd
import plotly.express as px
from transformers import pipeline

# Set the page layout for Streamlit
st.set_page_config(layout="wide")

# Initialize TAPAS pipeline for table-based question answering (multilingual)
tqa = pipeline(task="table-question-answering", 
              model="google/tapas-large-finetuned-wtq",
              device=0)  # Assuming GPU is available, otherwise set device="cpu"

# Title and Introduction
st.title("Data Visualization App with TAPAS NLP Integration")
st.markdown(""" 
    This app allows you to upload a table (CSV or Excel) and ask questions to generate graphs visualizing the data.
    Using **TAPAS**, the app can interpret your questions and generate the corresponding graphs.

    ### Available Features:
    - **Scatter Plot**: Visualize relationships between two columns.
    - **Line Graph**: Visualize a single column over time.
    
    Upload your data and ask questions about the data to generate visualizations.
""")

# Language Selection
language = st.selectbox(
    "Select the language of your question",
    ("English", "German", "French", "Spanish", "Italian", "Others")
)

# File uploader in the sidebar
file_name = st.sidebar.file_uploader("Upload file:", type=['csv', 'xlsx'])

# File processing and question answering
if file_name is None:
    st.markdown('<p class="font">Please upload an excel or csv file </p>', unsafe_allow_html=True)
else:
    try:
        # Check file type and handle reading accordingly
        if file_name.name.endswith('.csv'):
            df = pd.read_csv(file_name, sep=';', encoding='ISO-8859-1')  # Adjust encoding if needed
        elif file_name.name.endswith('.xlsx'):
            df = pd.read_excel(file_name, engine='openpyxl')  # Use openpyxl to read .xlsx files
        else:
            st.error("Unsupported file type")
            df = None

        if df is not None:
            # Convert object columns to numeric where possible
            df = df.apply(pd.to_numeric, errors='ignore')

            st.write("Original Data:")
            st.write(df)

            # Display a sample of data for graph generation
            st.write("Sample data for graph generation:")
            st.write(df.head())

    except Exception as e:
        st.error(f"Error reading file: {str(e)}")

    # User input for the question
    question = st.text_input(f'Ask your graph-related question in {language}')

    with st.spinner():
        if st.button('Generate Graph'):
            try:
                # Ensure the question is a string
                if not isinstance(question, str):
                    st.error("The question input is not a valid string.")
                    return

                # Use TAPAS model to process the question
                result = tqa(table=df, query=question)

                # Check if TAPAS is returning the expected answer
                answer = result['answer']
                st.write(f"TAPAS Answer: {answer}")

                # Determine if the question relates to graph generation
                if 'between' in question.lower() and 'and' in question.lower():
                    # This is a request for a scatter plot (two columns)
                    columns = question.split('between')[-1].split('and')
                    columns = [col.strip() for col in columns]
                    if len(columns) == 2 and all(col in df.columns for col in columns):
                        fig = px.scatter(df, x=columns[0], y=columns[1], title=f"Scatter Plot between {columns[0]} and {columns[1]}")
                        st.plotly_chart(fig, use_container_width=True)
                        st.success(f"Here is the scatter plot between '{columns[0]}' and '{columns[1]}'.")
                    else:
                        st.warning("Columns not found in the dataset.")
                elif 'column' in question.lower():
                    # This is a request for a line graph (single column)
                    column = question.split('of')[-1].strip()  # Handle 'of' keyword
                    if column in df.columns:
                        fig = px.line(df, x=df.index, y=column, title=f"Graph of column '{column}'")
                        st.plotly_chart(fig, use_container_width=True)
                        st.success(f"Here is the graph of column '{column}'.")
                    else:
                        st.warning(f"Column '{column}' not found in the data.")
                else:
                    st.warning("Please ask a valid graph-related question (e.g., 'make a graph between column1 and column2').")

            except Exception as e:
                st.warning(f"Error processing question or generating graph: {str(e)}")