import requests
from PIL import Image
from transformers import BlipProcessor, BlipForConditionalGeneration
import torch
from typing import Dict, List, Any

class EndpointHandler():
    def __init__(self, path=""):
        self.processor = BlipProcessor.from_pretrained("Salesforce/blip-image-captioning-large")
        self.model = BlipForConditionalGeneration.from_pretrained("Salesforce/blip-image-captioning-large")

        self.device = "cuda" if torch.cuda.is_available() else "cpu"
        self.model.to(self.device)

    def process_single_image(self, img_url, text=None):
        raw_image = Image.open(requests.get(img_url, stream=True).raw).convert('RGB')
        if text:
            inputs = self.processor(raw_image, text, return_tensors="pt").to(self.device)
        else:
            inputs = self.processor(raw_image, return_tensors="pt").to(self.device)

        out = self.model.generate(**inputs)
        return self.processor.decode(out[0], skip_special_tokens=True)

    def __call__(self, data: Dict[str, Any]) -> List[Dict[str, Any]]:
        try:
            print(f"Received data: {data}")

            if not data or "images" not in data:
                return [{"error": "No images data provided in the request."}]
            
            images_data = data.get("images")

            alt_texts = []
            for image in images_data:
                img_id = image.get("id")
                img_url = image.get("url")
                text = image.get("text", None)

                alt_text = self.process_single_image(img_url, text)
                alt_texts.append({
                    "image_id": img_id,
                    "image_url": img_url, 
                    "alt_text": alt_text
                })

            return alt_texts
        except Exception as e:
            print(f"Error processing data: {e}")
            return [{"error": str(e)}]

def get_pipeline(model_dir, task):
    return EndpointHandler(model_dir)