import numpy as np
from download_images import download_image_as_pil

def batch_process_images(image_urls: str, encoder, batch_size: int):
    """
    Processes a batch of image URLs and generates embeddings for each image.
    
    Args:
        image_urls (str): Comma-separated string of image URLs.
        encoder: The encoder object used to generate embeddings.
        batch_size (int): The maximum number of images to process in one batch.
    
    Returns:
        list: A list of dictionaries containing embedding results or errors.
    """
    # Split the input string by commas and strip whitespace
    urls = [url.strip() for url in image_urls.split(",") if url.strip()]
    
    if not urls:
        return {"error": "No valid image URLs provided."}

    results = []
    batch_urls, batch_images = [], []

    for url in urls:
        try:
            # Download image
            image = download_image_as_pil(url)
            if not image:
                results.append({"image_url": url, "error": "Failed to download image"})
                continue

            batch_urls.append(url)
            batch_images.append(image)

            # Process batch when reaching batch size
            if len(batch_images) == batch_size:
                process_batch(batch_urls, batch_images, results, encoder)
                batch_urls, batch_images = [], []

        except Exception as e:
            results.append({"image_url": url, "error": str(e)})

    # Process remaining images in the last batch
    if batch_images:
        process_batch(batch_urls, batch_images, results, encoder)

    return results


def process_batch(batch_urls, batch_images, results, encoder):
    """
    Processes a batch of images and generates embeddings.
    
    Args:
        batch_urls (list): List of image URLs in the batch.
        batch_images (list): List of PIL images in the batch.
        results (list): List to store results for each image.
        encoder: The encoder object used to generate embeddings.
    """
    try:
        # Generate embeddings
        embeddings = encoder.encode_images(batch_images)
        
        for url, embedding in zip(batch_urls, embeddings):
            # Normalize embedding
            embedding_normalized = embedding / np.linalg.norm(embedding)
            
            # Append results
            results.append({
                "image_url": url,
                "embedding_preview": embedding_normalized[:5].tolist(),  # First 5 values for preview
                "success": True
            })
    except Exception as e:
        for url in batch_urls:
            results.append({"image_url": url, "error": str(e)})