app.py

import gradio as gr
import pandas as pd
import os
import zipfile

def process_csv(uploaded_file):
    """
    Process the uploaded CSV file to:
    1. Replace text-based columns and numerical columns with less than six unique options with coded values.
    2. Fill missing values in numerical columns with their respective medians.
    3. Return a zip file containing the modified CSV file, a legend CSV, and a CSV detailing data fill methods.
    """
    # Load the data from the uploaded file's byte stream
    data = pd.read_csv(uploaded_file.name)
    
    # List to store mappings of columns
    mapping_list = []

    # List to store the details of columns where data was added
    data_added_details = []

    # Loop through each column in the DataFrame
    for col in data.columns:
        # Check if the column is of type object (text-based) or if it's numerical with less than six unique options
        if data[col].dtype == 'object' or (data[col].nunique() < 6 and pd.api.types.is_numeric_dtype(data[col])):
            # Create a mapping of original values to codes, including NaN or blank values mapped to -9999
            mapping = {value: code if pd.notna(value) else -9999 for code, value in enumerate(data[col].unique())}
            for original_value, mapped_value in mapping.items():
                mapping_list.append([col, original_value, mapped_value])
            # Replace the values in the column with their respective codes
            data[col] = data[col].map(mapping)
        elif pd.api.types.is_numeric_dtype(data[col]) and any(pd.isna(data[col])):
            # Replace with median
            median_value = data[col].median()
            data[col].fillna(median_value, inplace=True)
            data_added_details.append([col, "Median", median_value])

    # Name of the zip file
    zip_name = "processed_files.zip"

    # Save CSV files and add them to the zip file
    with zipfile.ZipFile(zip_name, 'w') as zipf:
        data.to_csv("modified_data.csv", index=False)
        zipf.write("modified_data.csv")

        mapping_df = pd.DataFrame(mapping_list, columns=['Column', 'Original Value', 'Mapped Value'])
        mapping_df.to_csv("mapping.csv", index=False)
        zipf.write("mapping.csv")

        data_added_df = pd.DataFrame(data_added_details, columns=['Column', 'Method', 'Value Added'])
        data_added_df.to_csv("data_added_details.csv", index=False)
        zipf.write("data_added_details.csv")

    return zip_name

# Gradio Interface
iface = gr.Interface(
    fn=process_csv,
    inputs=gr.inputs.File(type="file", label="Upload CSV File"),
    outputs=gr.outputs.File(label="Download Processed Files"),
    live=False
)

iface.launch()