app.py

import pandas as pd
from sentence_transformers import SentenceTransformer
import faiss
import numpy as np
import streamlit as st

# Load CSV data using a relative path
csv_file = "Hydra-Movie-Scrape.csv"  # Ensure this file is in the same directory as app.py
df = pd.read_csv(csv_file)

# Use 'Summary' or 'Short Summary' as the source for documents
# Fill NaNs with "No summary available."
df['Summary'] = df['Summary'].fillna("No summary available.")
documents = df['Summary'].tolist()  # Use 'Summary' for document embeddings

# Initialize the SentenceTransformer model
model = SentenceTransformer('all-MiniLM-L6-v2')

# Create and cache embeddings
@st.cache_resource
def create_embeddings(documents):
    embeddings = model.encode(documents, show_progress_bar=True)
    return embeddings

# Generate and cache embeddings
doc_embeddings = create_embeddings(documents)

# Convert to NumPy array (FAISS requires float32)
embedding_matrix = np.array(doc_embeddings).astype("float32")

# Build FAISS index for efficient similarity search
index = faiss.IndexFlatL2(embedding_matrix.shape[1])
index.add(embedding_matrix)

# Function to retrieve the most relevant documents
def retrieve(query, top_k=10):  # Retrieve up to 10 movies
    query_embedding = model.encode(query)  # Encode the query
    query_vector = np.array(query_embedding).astype("float32")
    distances, indices = index.search(np.array([query_vector]), top_k)
    return indices[0]

# Streamlit application layout
st.title("Movie Dataset RAG Application")
query = st.text_input("Ask a question about movies:")
if st.button("Submit"):
    if query:
        indices = retrieve(query)

        # Prepare and display detailed responses
        response = ""
        for idx in indices:
            if idx != -1:  # Check if the index is valid
                movie_details = df.iloc[idx]
                response += f"**Title**: {movie_details['Title']}\n"
                response += f"**Year**: {movie_details['Year']}\n"
                response += f"**Director**: {movie_details['Director']}\n"
                response += f"**Cast**: {movie_details['Cast']}\n"
                response += f"**Summary**: {movie_details['Summary']}\n\n"
        
        # Formatting the output with clearer separation
        if response:
            st.write("Here are some movies that match your query:\n")
            st.markdown(response)  # Use markdown to format the output nicely
        else:
            st.write("No relevant documents found.")
    else:
        st.write("Please enter a query.")