from typing import Any, Dict
from transformers import Blip2Processor, Blip2ForConditionalGeneration


class EndpointHandler:
    def __init__(self, path=""):
        # load model and processor from path
        self.processor = Blip2Processor.from_pretrained(path)
        self.model = Blip2ForConditionalGeneration.from_pretrained(path)
        self.device = "cuda"

        self.model.to(self.device)

    def __call__(self, data: Dict[str, Any]) -> Dict[str, str]:
        # process input
        image = data.pop("image", data)
        text = data.pop("text", data)

        inputs = self.processor(images=image, text=prompt, return_tensors="pt").to(self.device)
        generated_ids = self.model.generate(**inputs)
        generated_text = self.processor.batch_decode(generated_ids, skip_special_tokens=True)[0].strip()

        return [{"answer": generated_text}]