import pandas as pd
import gradio as gr
import matplotlib.pyplot as plt
import io
import base64
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.metrics.pairwise import linear_kernel

# Load your dataset
data = pd.read_csv('Blinkit Cart Prediction.csv')

# Create a TF-IDF vectorizer
tfidf_vectorizer = TfidfVectorizer(max_features=1000)  # Adjust max_features as needed
tfidf_matrix = tfidf_vectorizer.fit_transform(data['Description'])

# Function to recommend products
def recommend_products(user_choice, num_recommendations=10):
    # Transform the user choice using the same TF-IDF vectorizer
    user_choice_vector = tfidf_vectorizer.transform([user_choice])

    # Compute the cosine similarity between the user choice and all products
    cosine_similarities = linear_kernel(user_choice_vector, tfidf_matrix)

    # Get the indices of the most similar products
    similar_indices = cosine_similarities.argsort()[0][-num_recommendations - 1:-1][::-1]

    # Get the recommended product IDs and names
    recommended_products = data.iloc[similar_indices][['ProductID', 'ProductName']]

    return recommended_products

# Define the input and output components
input_component = gr.inputs.Textbox(label="Enter Your Choice")
output_component = gr.outputs.HTML(label="Recommended Products")

# Create the Gradio interface
def recommend_interface(user_choice):
    recommended_products = recommend_products(user_choice)
    
    # Create a bar graph of recommended products
    plt.figure(figsize=(10, 6))
    plt.bar(recommended_products['ProductName'], range(len(recommended_products)), color='skyblue')
    plt.xticks(rotation=45, ha="right")
    plt.xlabel("Recommended Products")
    plt.ylabel("Ranking")
    plt.title("Top Recommended Products")
    
    # Encode the image as base64
    buffer = io.BytesIO()
    plt.savefig(buffer, format="png")
    graph_base64 = base64.b64encode(buffer.getvalue()).decode()
    plt.close()
    
    # Create an HTML string with an embedded image
    graph_html = f'<img src="data:image/png;base64,{graph_base64}" />'
    
    # Create an HTML string for the table of recommended products
    table_html = recommended_products.to_html(index=False)
    
    # Concatenate the HTML strings for both components
    result_html = f"<h2>Recommended Products:</h2>{table_html}<br>{graph_html}"
    
    return result_html

# Launch the Gradio app
interface = gr.Interface(
fn=recommend_interface, inputs=input_component, outputs=output_component,
            live = True,   
    description = " Press flag if any erroneous output comes ",
    theme=gr.themes.Soft(),
    title = "Blinkit Cart Prediction",
    examples = [['necklace'],
                ['DSLR camera '],['tea '], ['Smart TV '] , ['protein bars'] , ['sunglasses ']  ],
    )
interface.launch(inline=False)