import oneclass
import Cosine_distance
import gradio as gr
import pandas as pd
from io import StringIO

def predict_and_download(classifier_oneclass_or_Cosine_distance ,positive_csv_file, unlabelled_csv_file, n,Hyperparameter_nu):
    if classifier_oneclass_or_Cosine_distance == "oneclass" :
       selected_paper_info = oneclass.select_top_n_papers(n, positive_csv_file, unlabelled_csv_file,Hyperparameter_nu)
    else :    
       selected_paper_info = Cosine_distance.recommend_papers(positive_csv_file, positive_csv_file, n)
        
    # Create a StringIO object to store CSV data
    csv_buffer = StringIO()
    
    # Write DataFrame to the StringIO buffer as CSV
    selected_paper_info.to_csv(csv_buffer, index=False)
    
    # Get the CSV data from the buffer
    csv_content = csv_buffer.getvalue()
    
    
    # Return selected_paper_info and CSV content
    return selected_paper_info, csv_buffer

# Create the interface
iface = gr.Interface(
    fn=predict_and_download,
    inputs=["text","file", "file", "number"],
    outputs=[gr.DataFrame(label="Recommended Papers"), gr.DownloadButton(label="Download CSV")],
    title="Personalized Arxiv Feed",
    description="Enter text and upload CSV files for labelled and unlabelled data.",
    allow_flagging='never'  # Disable flagging feature
)

iface.launch()