app.py

# app.py

import gradio as gr
from PIL import Image
import base64
import requests
import os
from io import BytesIO

from utils.planner import extract_scene_plan  # 🧠 Brain Layer

# 🔐 Hugging Face keys
HF_API_KEY = os.getenv("HF_API_KEY")
SDXL_MODEL_ID = "fal-ai/fast-sdxl/image-to-image"  # ✅ Correct model for image-to-image
SDXL_API_URL = f"https://api-inference.huggingface.co/models/{SDXL_MODEL_ID}"
HEADERS = {"Authorization": f"Bearer {HF_API_KEY}"}

# 🚀 Image generation (img2img)
def process_image(prompt, image, num_variations):
    try:
        print("🧠 Prompt received:", prompt)
        
        # Step 1: Brain Layer
        reasoning_json = extract_scene_plan(prompt)
        print("🧠 Scene plan extracted:", reasoning_json)

        # Step 2: Encode input image
        buffered = BytesIO()
        image.save(buffered, format="JPEG")
        img_bytes = buffered.getvalue()
        encoded_image = base64.b64encode(img_bytes).decode("utf-8")

        # Step 3: Send image + prompt to HF API
        outputs = []
        for i in range(num_variations):
            payload = {
                "image": encoded_image,
                "prompt": prompt,
                "negative_prompt": "blurry, deformed, cropped",
                "num_inference_steps": 25,
                "guidance_scale": 7.5
            }

            print(f"📤 Sending request to HF (variation {i+1})")
            response = requests.post(SDXL_API_URL, headers=HEADERS, json=payload)

            if response.status_code == 200:
                try:
                    result_json = response.json()
                    if "images" in result_json:
                        base64_img = result_json["images"][0]
                        result_image = Image.open(BytesIO(base64.b64decode(base64_img)))
                        outputs.append(result_image)
                        print(f"✅ Decoded image variation {i+1} successfully")
                    else:
                        print(f"⚠️ No 'images' key found in response")
                        outputs.append("❌ No image in response.")
                except Exception as decode_err:
                    print("❌ Image decode error:", decode_err)
                    outputs.append("❌ Failed to decode image.")
            else:
                print(f"❌ HF API error: {response.status_code} - {response.text}")
                outputs.append(f"Error {response.status_code}: {response.text}")

        return outputs, reasoning_json

    except Exception as e:
        print("❌ General Exception in process_image:", e)
        return ["Processing error occurred"], {"error": str(e)}

# 🎨 Gradio UI
with gr.Blocks() as demo:
    gr.Markdown("# 🧠 NewCrux AI Demo: Image-to-Image using Fast SDXL + Brain Layer")

    with gr.Row():
        with gr.Column():
            prompt_input = gr.Textbox(label="Enter Prompt")
            image_input = gr.Image(type="pil", label="Upload Product Image")
            variation_slider = gr.Slider(1, 4, step=1, value=1, label="Number of Variations")
            generate_btn = gr.Button("Generate")

        with gr.Column():
            output_gallery = gr.Gallery(
                label="Generated Image Variations",
                columns=2,
                rows=2,
                height="auto"
            )
            json_output = gr.JSON(label="Brain Layer Reasoning (Scene Plan)")

    generate_btn.click(
        fn=process_image,
        inputs=[prompt_input, image_input, variation_slider],
        outputs=[output_gallery, json_output]
    )

demo.launch(share=True)