app.py

import torch
from transformers import AutoTokenizer, AutoModelForCausalLM
import gradio as gr
import requests

model_name = "Writer/palmyra-small"
tokenizer = AutoTokenizer.from_pretrained(model_name)
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model = AutoModelForCausalLM.from_pretrained(model_name).to(device)

def get_movie_info(movie_title):
    api_key = "20e959f0f28e6b3e3de49c50f358538a"
    search_url = f"https://api.themoviedb.org/3/search/movie"
    
    # Make a search query to TMDb
    params = {
        "api_key": api_key,
        "query": movie_title,
        "language": "en-US",
        "page": 1,
    }

    try:
        search_response = requests.get(search_url, params=params)
        search_data = search_response.json()

        # Check if any results are found
        if search_data.get("results"):
            movie_id = search_data["results"][0]["id"]

            # Fetch detailed information using the movie ID
            details_url = f"https://api.themoviedb.org/3/movie/{movie_id}"
            details_params = {
                "api_key": api_key,
                "language": "en-US",
            }

            details_response = requests.get(details_url, params=details_params)
            details_data = details_response.json()

            # Extract relevant information
            title = details_data.get("title", "Unknown Title")
            year = details_data.get("release_date", "Unknown Year")[:4]
            genre = ", ".join(genre["name"] for genre in details_data.get("genres", []))
            tmdb_link = f"https://www.themoviedb.org/movie/{movie_id}"
            poster_path = details_data.get("poster_path")
            
            # Convert poster_path to a complete image URL
            image_url = f"https://image.tmdb.org/t/p/w500{poster_path}" if poster_path else ""

            return f"Title: {title}, Year: {year}, Genre: {genre}\nFind more info here: {tmdb_link}", image_url

        else:
            return "Movie not found", ""

    except Exception as e:
        return f"Error: {e}", ""

def generate_response(prompt):
    input_text_template = (
        "A chat between a curious user and an artificial intelligence assistant. "
        "The assistant gives helpful, detailed, and polite answers to the user's questions. "
        f"USER: {prompt} "
        "ASSISTANT:"
    )

    # Call the get_movie_info function to enrich the response
    movie_info, image_url = get_movie_info(prompt)

    # Concatenate the movie info with the input template
    input_text_template += f" Movie Info: {movie_info}"

    model_inputs = tokenizer(input_text_template, return_tensors="pt").to(device)

    gen_conf = {
        "top_k": 20,
        "max_length": 20,
        "temperature": 0.6,
        "do_sample": True,
        "eos_token_id": tokenizer.eos_token_id,
    }

    output = model.generate(**model_inputs, **gen_conf)

    generated_text = tokenizer.decode(output[0], skip_special_tokens=True)
    
    # Display image directly in the chat
    image_component = gr.Image(image_url, alt="Movie Poster")
    
    return f"Movie Info:\n{movie_info}\n\nGenerated Response:\n{generated_text}\n", image_component

# Define chat function for gr.ChatInterface
def chat_function(message, history):
    response, image_component = generate_response(message)
    history.append([message, response])
    return response, image_component

# Create Gradio Chat Interface
chat_interface = gr.ChatInterface(chat_function)
chat_interface.launch()