app.py

import gradio as gr
from transformers import AutoTokenizer, AutoModelForQuestionAnswering
import torch
import logging
import warnings
from typing import List, Tuple, Dict
import random
import hashlib
from datetime import datetime
from dataclasses import dataclass
from enum import Enum
import json
import re
from pathlib import Path

logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)
warnings.filterwarnings('ignore')

class ThemeType(Enum):
    MARRIAGE = "casamento"
    FAMILY = "familia"
    SPIRITUAL = "vida_espiritual"
    WORK = "trabalho"
    RELATIONSHIPS = "relacionamentos"
    GENERAL = "geral"

@dataclass
class BiblicalExample:
    question: str
    passage: str
    text: str
    base_response: str
    application: str
    keywords: List[str]
    sentiment: str  # Added for emotional context

class ResponseGenerator:
    def __init__(self):
        self.nlp_model = None  # Placeholder for sentiment analysis
        
    def analyze_sentiment(self, text: str) -> str:
        # Basic sentiment analysis based on keywords
        positive_words = {'alegria', 'esperança', 'paz', 'amor', 'gratidão'}
        negative_words = {'tristeza', 'medo', 'ansiedade', 'preocupação', 'angústia'}
        
        text_words = set(text.lower().split())
        pos_count = len(text_words.intersection(positive_words))
        neg_count = len(text_words.intersection(negative_words))
        
        return 'positive' if pos_count > neg_count else 'negative' if neg_count > pos_count else 'neutral'

    def personalize_response(self, example: BiblicalExample, question: str) -> str:
        sentiment = self.analyze_sentiment(question)
        
        # Customize response based on sentiment
        intro = {
            'positive': "Que bom que você está buscando orientação! ",
            'negative': "Entendo que você possa estar passando por um momento difícil. ",
            'neutral': "Agradeço sua busca por sabedoria. "
        }
        
        return f"{intro[sentiment]}{example.base_response}"

class EnhancedBiblicalCounselor:
    def __init__(self):
        logger.info("Inicializando conselheiro bíblico aprimorado...")
        self.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
        self.model_name = "pierreguillou/bert-base-cased-squad-v1.1-portuguese"
        self.session_history = []
        self.response_generator = ResponseGenerator()
        self.load_examples()
        self.setup_model()
        
    def load_examples(self):
        # Load examples from JSON file (you would need to create this)
        examples_path = Path("biblical_examples.json")
        if examples_path.exists():
            with open(examples_path) as f:
                self.biblical_examples = json.load(f)
        else:
            # Fallback to default examples
            self.biblical_examples = self._get_default_examples()
    
    def setup_model(self):
        try:
            self.tokenizer = AutoTokenizer.from_pretrained(self.model_name)
            self.model = AutoModelForQuestionAnswering.from_pretrained(self.model_name)
            self.model.to(self.device)
            logger.info(f"Modelo carregado com sucesso no dispositivo: {self.device}")
        except Exception as e:
            logger.error(f"Erro ao carregar modelo: {str(e)}")
            raise

    def get_unique_response(self, question: str, theme: str = None) -> Tuple[str, Dict, str]:
        if not theme:
            theme = self.find_best_theme(question)
        
        example = self._select_best_example(question, theme)
        personalized_response = self.response_generator.personalize_response(example, question)
        
        # Generate AI response using the model
        model_response = self._generate_model_response(question, example['texto'])
        
        # Create enhanced response with better formatting and structure
        response = self._format_enhanced_response(
            personalized_response,
            example,
            model_response,
            self._get_additional_resources(theme)
        )
        
        metadata = self._create_metadata(example, theme)
        history = self.save_to_history(question, theme, response, metadata)
        
        return response, metadata, history

    def _format_enhanced_response(self, personalized_response, example, model_response, resources):
        return f"""
🌟 Orientação Personalizada:
{personalized_response}

📖 Passagem Bíblica:
{example['passagem']}: {example['texto']}

✨ Aplicação Prática:
{example['aplicacao']}

💭 Reflexão Gerada por IA:
{model_response}

📚 Recursos Adicionais:
{resources}

🙏 Observação: Esta orientação é baseada em princípios bíblicos. Para questões específicas, 
considere consultar sua liderança espiritual local.
"""

    def _get_additional_resources(self, theme):
        # Add relevant books, articles, or other resources based on the theme
        resources = {
            "casamento": ["'O Significado do Casamento' - Timothy Keller", 
                         "'Casamento Blindado' - Renato e Cristiane Cardoso"],
            "familia": ["'Criando Filhos' - Tim Kimmel",
                       "'Limites' - Henry Cloud e John Townsend"]
            # Add more resources for other themes
        }
        return "\n".join(resources.get(theme, ["Bíblia de Estudo"]))

def create_enhanced_interface():
    counselor = EnhancedBiblicalCounselor()
    
    custom_theme = gr.themes.Soft().set(
        button_primary_background_fill="indigo",
        button_primary_background_fill_dark="darkblue",
    )

    with gr.Blocks(theme=custom_theme) as demo:
        gr.Markdown("""
        # 🕊️ Conselheiro Bíblico Plus
        ### Orientação Bíblica Personalizada com Inteligência Artificial
        """)

        with gr.Row():
            with gr.Column(scale=1):
                verse_of_day = gr.Textbox(
                    label="🌟 Versículo do Dia",
                    value=counselor.get_verse_of_the_day(),
                    lines=4,
                    interactive=False
                )
            
            with gr.Column(scale=1):
                prayer_focus = gr.Textbox(
                    label="🙏 Foco de Oração do Dia",
                    value=counselor.get_daily_prayer_focus(),
                    lines=4,
                    interactive=False
                )

        with gr.Tabs():
            with gr.TabItem("📝 Nova Consulta", id=1):
                with gr.Row():
                    with gr.Column():
                        theme = gr.Dropdown(
                            choices=counselor.get_themes(),
                            label="🎯 Tema",
                            value="geral",
                            info="Selecione um tema ou deixe em automático"
                        )
                        
                        question = gr.Textbox(
                            label="❓ Sua Pergunta",
                            placeholder="Digite sua pergunta sobre qualquer tema bíblico...",
                            lines=3
                        )
                        
                        with gr.Row():
                            submit_btn = gr.Button("🙏 Buscar Orientação", variant="primary")
                            clear_btn = gr.Button("🔄 Limpar", variant="secondary")

                    with gr.Column():
                        answer_output = gr.Textbox(
                            label="✨ Orientação",
                            lines=12
                        )
                        
                        with gr.Accordion("📚 Detalhes e Referências"):
                            metadata_output = gr.JSON(
                                label="📋 Informações Detalhadas"
                            )
                            
                        feedback = gr.Radio(
                            ["👍 Útil", "👎 Precisa Melhorar"],
                            label="📢 Sua Opinião"
                        )

            # Add more tabs and features...

    return demo

if __name__ == "__main__":
    try:
        logger.info("Iniciando aplicação aprimorada...")
        demo = create_enhanced_interface()
        demo.launch(
            server_name="0.0.0.0",
            share=True,
            show_error=True,
            server_port=7860
        )
    except Exception as e:
        logger.error(f"Erro ao iniciar aplicação: {str(e)}")
        raise