import os
import sys
import json
import torch
from ts.torch_handler.base_handler import BaseHandler
from transformers import BertTokenizer, BertForSequenceClassification

# Add the model directory to the Python path
model_dir = os.path.dirname(os.path.abspath(__file__))
sys.path.append(model_dir)

from model import ImprovedBERTClass  # Ensure this import matches your model file name

class UICardMappingHandler(BaseHandler):
    def __init__(self):
        super().__init__()
        self.initialized = False

    def initialize(self, context):
        self.manifest = context.manifest
        properties = context.system_properties
        model_dir = properties.get("model_dir")
        self.device = torch.device("cuda:" + str(properties.get("gpu_id")) if torch.cuda.is_available() else "cpu")

        self.tokenizer = BertTokenizer.from_pretrained(model_dir)
        self.model = ImprovedBERTClass()
        self.model.load_state_dict(torch.load(os.path.join(model_dir, 'model.pth'), map_location=self.device))
        self.model.to(self.device)
        self.model.eval()

        self.initialized = True

    def preprocess(self, data):
        text = data[0].get("data")
        if text is None:
            text = data[0].get("body")
        inputs = self.tokenizer(text, return_tensors="pt", max_length=64, padding='max_length', truncation=True)
        return inputs.to(self.device)

    def inference(self, inputs):
        with torch.no_grad():
            outputs = self.model(**inputs)
        return torch.sigmoid(outputs.logits)

    def postprocess(self, inference_output):
        probabilities = inference_output.cpu().numpy().flatten()
        labels = ['Videos', 'Unit Conversion', 'Translation', 'Shopping Product Comparison', 'Restaurants', 'Product', 'Information', 'Images', 'Gift', 'General Comparison', 'Flights', 'Answer', 'Aircraft Seat Map']
        
        top_k = 3  # You can adjust this value
        top_k_indices = probabilities.argsort()[-top_k:][::-1]
        top_k_probs = probabilities[top_k_indices]
        
        top_k_predictions = [{"card": labels[i], "probability": float(p)} for i, p in zip(top_k_indices, top_k_probs)]
        
        most_likely_card = "Answer" if sum(probabilities > 0.5) == 0 else labels[probabilities.argmax()]
        
        result = {
            "most_likely_card": most_likely_card,
            "top_k_predictions": top_k_predictions
        }
        
        return [result]