import asyncio
from generate_prompts import generate_prompt
from diffusers import AutoPipelineForText2Image
from io import BytesIO
import gradio as gr
import anyio

# Asynchronously load the model once outside of the function
model = None

async def load_model():
    global model
    print("Loading the model...")
    model = AutoPipelineForText2Image.from_pretrained("stabilityai/sdxl-turbo")
    print("Model loaded successfully.")

# Run the model loading
asyncio.run(load_model())

async def generate_image(prompt, prompt_name):
    try:
        print(f"Generating response for {prompt_name} with prompt: {prompt}")
        output = model(prompt=prompt, num_inference_steps=1, guidance_scale=0.0)
        print(f"Output for {prompt_name}: {output}")

        # Check if the model returned images
        if isinstance(output.images, list) and len(output.images) > 0:
            image = output.images[0]
            buffered = BytesIO()
            try:
                image.save(buffered, format="JPEG")
                image_bytes = buffered.getvalue()
                print(f"Image bytes length for {prompt_name}: {len(image_bytes)}")
                return image_bytes
            except Exception as e:
                print(f"Error saving image for {prompt_name}: {e}")
                return None
        else:
            raise Exception(f"No images returned by the model for {prompt_name}.")
    except Exception as e:
        print(f"Error generating image for {prompt_name}: {e}")
        return None

async def queue_api_calls(sentence_mapping, character_dict, selected_style):
    print(f"queue_api_calls invoked with sentence_mapping: {sentence_mapping}, character_dict: {character_dict}, selected_style: {selected_style}")
    prompts = []

    # Generate prompts for each paragraph
    for paragraph_number, sentences in sentence_mapping.items():
        combined_sentence = " ".join(sentences)
        print(f"combined_sentence for paragraph {paragraph_number}: {combined_sentence}")
        prompt = generate_prompt(combined_sentence, character_dict, selected_style)  # Correct prompt generation
        prompts.append((paragraph_number, prompt))
        print(f"Generated prompt for paragraph {paragraph_number}: {prompt}")

    # Generate images for each prompt in parallel
    tasks = [generate_image(prompt, f"Prompt {paragraph_number}") for paragraph_number, prompt in prompts]
    print("Tasks created for image generation.")
    responses = await asyncio.gather(*tasks)
    print("Responses received from image generation tasks.")

    images = {paragraph_number: response for (paragraph_number, _), response in zip(prompts, responses)}
    print(f"Images generated: {images}")
    return images

def process_prompt(sentence_mapping, character_dict, selected_style):
    print(f"process_prompt called with sentence_mapping: {sentence_mapping}, character_dict: {character_dict}, selected_style: {selected_style}")
    
    async def run_async():
        async with anyio.create_task_group() as task_group:
            return await task_group.start(queue_api_calls, sentence_mapping, character_dict, selected_style)
    
    cmpt_return = anyio.run(run_async)
    print(f"process_prompt completed with return value: {cmpt_return}")
    return cmpt_return

# Gradio interface with high concurrency limit
gradio_interface = gr.Interface(
    fn=process_prompt,
    inputs=[
        gr.JSON(label="Sentence Mapping"),
        gr.JSON(label="Character Dict"),
        gr.Dropdown(["oil painting", "sketch", "watercolor"], label="Selected Style")
    ],
    outputs="json"
)

if __name__ == "__main__":
    print("Launching Gradio interface...")
    gradio_interface.launch()
    print("Gradio interface launched.")