import pandas as pd
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.metrics.pairwise import linear_kernel
import gradio as gr

# Sample medicine data
data = {
    'drug_name': ['Paracetamol', 'Ibuprofen', 'Aspirin', 'Amoxicillin', 'Ciprofloxacin', 'Lisinopril'],
    'composition': [
        'Paracetamol 500mg',
        'Ibuprofen 200mg',
        'Aspirin 100mg',
        'Amoxicillin 500mg',
        'Ciprofloxacin 500mg',
        'Lisinopril 10mg'
    ],
    'description': [
        'Used for pain relief and fever reduction.',
        'Nonsteroidal anti-inflammatory drug (NSAID).',
        'Used to reduce pain, fever, or inflammation.',
        'Antibiotic used to treat bacterial infections.',
        'Antibiotic used to treat a variety of bacterial infections.',
        'Used to treat high blood pressure.'
    ]
}

# Create a DataFrame from the sample data
medicines_df = pd.DataFrame(data)

# Create the TF-IDF matrix for finding similar medicines
tfidf = TfidfVectorizer(stop_words='english')
tfidf_matrix = tfidf.fit_transform(medicines_df['composition'].fillna(''))

# Compute the cosine similarity matrix
cosine_sim = linear_kernel(tfidf_matrix, tfidf_matrix)

# Function to get recommendations based on drug name or composition
def get_alternatives(drug_name, cosine_sim=cosine_sim):
    if drug_name not in medicines_df['drug_name'].values:
        return pd.DataFrame(columns=['drug_name', 'composition'])  # Return empty DataFrame if not found
    idx = medicines_df.index[medicines_df['drug_name'] == drug_name][0]
    sim_scores = list(enumerate(cosine_sim[idx]))
    sim_scores = sorted(sim_scores, key=lambda x: x[1], reverse=True)
    sim_scores = sim_scores[1:4]  # Get top 3 alternatives
    alternative_indices = [i[0] for i in sim_scores]
    return medicines_df.iloc[alternative_indices][['drug_name', 'composition']]

# Define the recommend_alternative function for Gradio
def recommend_alternative(selected_drug):
    alternatives = get_alternatives(selected_drug)
    return alternatives

# Create the Gradio interface
drug_names = medicines_df['drug_name'].dropna().tolist()  # List of drug names
interface = gr.Interface(
    fn=recommend_alternative,
    inputs=gr.Dropdown(choices=drug_names, label="Select a Drug"),
    outputs="dataframe",
    title="Medicine Alternative Recommendation System",
    description="Select a medicine to see alternative drugs with similar compositions."
)

# Launch the interface
if __name__ == "__main__":
    interface.launch()