app.py

# import logging
# import random
# import warnings
# import gradio as gr
# import os
# import shutil
# import subprocess
# import spaces
# import torch
# import numpy as np
# from diffusers import FluxControlNetModel
# from diffusers.pipelines import FluxControlNetPipeline
# from PIL import Image
# from huggingface_hub import snapshot_download, login
# import io
# import base64
# from flask import Flask, request, jsonify
# from concurrent.futures import ThreadPoolExecutor
# from flask_cors import CORS
# import threading

# # Configure logging
# logging.basicConfig(level=logging.INFO)
# logger = logging.getLogger(__name__)

# app = Flask(__name__)
# CORS(app)

# # Function to check disk usage
# def check_disk_space():
#     result = subprocess.run(['df', '-h'], capture_output=True, text=True)
#     logger.info("Disk space usage:\n%s", result.stdout)

# # Function to clear Hugging Face cache
# def clear_huggingface_cache():
#     cache_dir = os.path.expanduser('~/.cache/huggingface')
#     if os.path.exists(cache_dir):
#         shutil.rmtree(cache_dir)  # Removes the entire cache directory
#         logger.info("Cleared Hugging Face cache at: %s", cache_dir)
#     else:
#         logger.info("No Hugging Face cache found.")

# # Check disk space
# check_disk_space()

# # Clear Hugging Face cache
# clear_huggingface_cache()

# # Add config to store base64 images
# app.config['image_outputs'] = {}

# # ThreadPoolExecutor for managing image processing threads
# executor = ThreadPoolExecutor()

# # Determine the device (GPU or CPU)
# if torch.cuda.is_available():
#     device = "cuda"
#     logger.info("CUDA is available. Using GPU.")
# else:
#     device = "cpu"
#     logger.info("CUDA is not available. Using CPU.")

# # Load model from Huggingface Hub
# huggingface_token = os.getenv("HUGGINGFACE_TOKEN")
# if huggingface_token:
#     login(token=huggingface_token)
#     logger.info("Hugging Face token found and logged in.")
# else:
#     logger.warning("Hugging Face token not found in environment variables.")

# logger.info("Hugging Face token: %s", huggingface_token)

# # Download model using snapshot_download

# model_path = snapshot_download(
#         repo_id="black-forest-labs/FLUX.1-dev", 
#         repo_type="model", 
#         ignore_patterns=["*.md", "*..gitattributes"],
#         local_dir="FLUX.1-dev",
#         token=huggingface_token)
# logger.info("Model downloaded to: %s", model_path)

# # Load pipeline
# logger.info('Loading ControlNet model.')

# controlnet = FluxControlNetModel.from_pretrained(
#         "jasperai/Flux.1-dev-Controlnet-Upscaler", torch_dtype=torch.bfloat16
#     ).to(device)
# logger.info("ControlNet model loaded successfully.")

# logger.info('Loading pipeline.')

# pipe = FluxControlNetPipeline.from_pretrained(
#         model_path, controlnet=controlnet, torch_dtype=torch.bfloat16
#     ).to(device)
# logger.info("Pipeline loaded successfully.")

# MAX_SEED = 1000000
# MAX_PIXEL_BUDGET = 1024 * 1024


# @spaces.GPU
# def process_input(input_image, upscale_factor):
#     w, h = input_image.size
#     aspect_ratio = w / h
#     was_resized = False

#     # Resize if input size exceeds the maximum pixel budget
#     if w * h * upscale_factor**2 > MAX_PIXEL_BUDGET:
#         warnings.warn(f"Requested output image is too large. Resizing to fit within pixel budget.")
#         input_image = input_image.resize(
#             (
#                 int(aspect_ratio * MAX_PIXEL_BUDGET**0.5 // upscale_factor),
#                 int(MAX_PIXEL_BUDGET**0.5 // aspect_ratio // upscale_factor),
#             )
#         )
#         was_resized = True

#     # Adjust dimensions to be a multiple of 8
#     w, h = input_image.size
#     w = w - w % 8
#     h = h - h % 8

#     return input_image.resize((w, h)), was_resized

# @spaces.GPU
# def run_inference(process_id, input_image, upscale_factor, seed, num_inference_steps, controlnet_conditioning_scale):
#     logger.info("Processing inference for process_id: %s", process_id)
#     input_image, was_resized = process_input(input_image, upscale_factor)

#     # Rescale image for ControlNet processing
#     w, h = input_image.size
#     control_image = input_image.resize((w * upscale_factor, h * upscale_factor))

#     # Set the random generator for inference
#     generator = torch.Generator().manual_seed(seed)

#     # Perform inference using the pipeline
#     logger.info("Running pipeline for process_id: %s", process_id)
#     image = pipe(
#         prompt="",
#         control_image=control_image,
#         controlnet_conditioning_scale=controlnet_conditioning_scale,
#         num_inference_steps=num_inference_steps,
#         guidance_scale=3.5,
#         height=control_image.size[1],
#         width=control_image.size[0],
#         generator=generator,
#     ).images[0]

#     # Resize output image back to the original dimensions if needed
#     if was_resized:
#         original_size = (input_image.width * upscale_factor, input_image.height * upscale_factor)
#         image = image.resize(original_size)

#     # Convert the output image to base64
#     buffered = io.BytesIO()
#     image.save(buffered, format="JPEG")
#     image_base64 = base64.b64encode(buffered.getvalue()).decode("utf-8")

#     # Store the result in the shared dictionary
#     app.config['image_outputs'][process_id] = image_base64
#     logger.info("Inference completed for process_id: %s", process_id)

# @app.route('/infer', methods=['POST'])
# def infer():
#     # Check if the file was provided in the form-data
#     if 'input_image' not in request.files:
#         logger.error("No image file provided in request.")
#         return jsonify({
#             "status": "error",
#             "message": "No input_image file provided"
#         }), 400

#     # Get the uploaded image file from the request
#     file = request.files['input_image']

#     # Check if a file was uploaded
#     if file.filename == '':
#         logger.error("No selected file in form-data.")
#         return jsonify({
#             "status": "error",
#             "message": "No selected file"
#         }), 400

#     # Convert the image to Base64 for internal processing
#     input_image = Image.open(file)
#     buffered = io.BytesIO()
#     input_image.save(buffered, format="JPEG")

#     # Retrieve additional parameters from the request (if any)
#     seed = request.form.get("seed", 42, type=int)
#     randomize_seed = request.form.get("randomize_seed", 'true').lower() == 'true'
#     num_inference_steps = request.form.get("num_inference_steps", 28, type=int)
#     upscale_factor = request.form.get("upscale_factor", 4, type=int)
#     controlnet_conditioning_scale = request.form.get("controlnet_conditioning_scale", 0.6, type=float)

#     # Randomize seed if specified
#     if randomize_seed:
#         seed = random.randint(0, MAX_SEED)
#         logger.info("Seed randomized to: %d", seed)

#     # Create a unique process ID for this request
#     process_id = str(random.randint(1000, 9999))
#     logger.info("Process started with process_id: %s", process_id)

#     # Set the status to 'in_progress'
#     app.config['image_outputs'][process_id] = None

#     # Run the inference in a separate thread
#     executor.submit(run_inference, process_id, input_image, upscale_factor, seed, num_inference_steps, controlnet_conditioning_scale)

#     # Return the process ID
#     return jsonify({
#         "process_id": process_id,
#         "message": "Processing started"
#     })


# # Modify status endpoint to receive process_id in request body
# @app.route('/status', methods=['GET'])
# def status():
#     # Get the process_id from the query parameters
#     process_id = request.args.get('process_id')

#     # Check if process_id was provided
#     if not process_id:
#         logger.error("Process ID not provided in request.")
#         return jsonify({
#             "status": "error",
#             "message": "Process ID is required"
#         }), 400

#     # Check if the process_id exists in the dictionary
#     if process_id not in app.config['image_outputs']:
#         logger.error("Invalid process ID: %s", process_id)
#         return jsonify({
#             "status": "error",
#             "message": "Invalid process ID"
#         }), 404

#     # Check the status of the image processing
#     image_base64 = app.config['image_outputs'][process_id]
#     if image_base64 is None:
#         logger.info("Process ID %s is still in progress.", process_id)
#         return jsonify({
#             "status": "in_progress"
#         })
#     else:
#         logger.info("Process ID %s completed successfully.", process_id)
#         return jsonify({
#             "status": "completed",
#             "output_image": image_base64
#         })


# if __name__ == '__main__':
#     app.run(debug=True,host='0.0.0.0')





# import logging
# import random
# import warnings
# import gradio as gr
# import os
# import shutil
# import subprocess
# import torch
# import numpy as np
# from diffusers import FluxControlNetModel
# from diffusers.pipelines import FluxControlNetPipeline
# from PIL import Image
# from huggingface_hub import snapshot_download, login
# import io
# import base64
# import threading

# # Configure logging
# logging.basicConfig(level=logging.INFO)
# logger = logging.getLogger(__name__)

# # Function to check disk usage
# def check_disk_space():
#     result = subprocess.run(['df', '-h'], capture_output=True, text=True)
#     logger.info("Disk space usage:\n%s", result.stdout)

# # Function to clear Hugging Face cache
# def clear_huggingface_cache():
#     cache_dir = os.path.expanduser('~/.cache/huggingface')
#     if os.path.exists(cache_dir):
#         shutil.rmtree(cache_dir)  # Removes the entire cache directory
#         logger.info("Cleared Hugging Face cache at: %s", cache_dir)
#     else:
#         logger.info("No Hugging Face cache found.")

# # Check disk space
# check_disk_space()

# # Clear Hugging Face cache
# clear_huggingface_cache()

# # Add config to store base64 images
# image_outputs = {}

# # Determine the device (GPU or CPU)
# if torch.cuda.is_available():
#     device = "cuda"
#     logger.info("CUDA is available. Using GPU.")
# else:
#     device = "cpu"
#     logger.info("CUDA is not available. Using CPU.")

# # Load model from Huggingface Hub
# huggingface_token = os.getenv("HUGGINGFACE_TOKEN")
# if huggingface_token:
#     login(token=huggingface_token)
#     logger.info("Hugging Face token found and logged in.")
# else:
#     logger.warning("Hugging Face token not found in environment variables.")

# # Download model using snapshot_download
# model_path = snapshot_download(
#     repo_id="black-forest-labs/FLUX.1-dev", 
#     repo_type="model", 
#     ignore_patterns=["*.md", "*..gitattributes"],
#     local_dir="FLUX.1-dev",
#     token=huggingface_token
# )
# logger.info("Model downloaded to: %s", model_path)

# # Load pipeline
# logger.info('Loading ControlNet model.')
# controlnet = FluxControlNetModel.from_pretrained(
#     "jasperai/Flux.1-dev-Controlnet-Upscaler", torch_dtype=torch.bfloat16
# ).to(device)
# logger.info("ControlNet model loaded successfully.")

# logger.info('Loading pipeline.')
# pipe = FluxControlNetPipeline.from_pretrained(
#     model_path, controlnet=controlnet, torch_dtype=torch.bfloat16
# ).to(device)
# logger.info("Pipeline loaded successfully.")

# MAX_SEED = 1000000
# MAX_PIXEL_BUDGET = 1024 * 1024

# def process_input(input_image, upscale_factor):
#     w, h = input_image.size
#     aspect_ratio = w / h
#     was_resized = False

#     # Resize if input size exceeds the maximum pixel budget
#     if w * h * upscale_factor**2 > MAX_PIXEL_BUDGET:
#         warnings.warn(f"Requested output image is too large. Resizing to fit within pixel budget.")
#         input_image = input_image.resize(
#             (
#                 int(aspect_ratio * MAX_PIXEL_BUDGET**0.5 // upscale_factor),
#                 int(MAX_PIXEL_BUDGET**0.5 // aspect_ratio // upscale_factor),
#             )
#         )
#         was_resized = True

#     # Adjust dimensions to be a multiple of 8
#     w, h = input_image.size
#     w = w - w % 8
#     h = h - h % 8

#     return input_image.resize((w, h)), was_resized

# def run_inference(input_image, upscale_factor, seed, num_inference_steps, controlnet_conditioning_scale):
#     logger.info("Running inference")
#     input_image, was_resized = process_input(input_image, upscale_factor)

#     # Rescale image for ControlNet processing
#     w, h = input_image.size
#     control_image = input_image.resize((w * upscale_factor, h * upscale_factor))

#     # Set the random generator for inference
#     generator = torch.Generator().manual_seed(seed)

#     # Perform inference using the pipeline
#     logger.info("Running pipeline")
#     image = pipe(
#         prompt="",
#         control_image=control_image,
#         controlnet_conditioning_scale=controlnet_conditioning_scale,
#         num_inference_steps=num_inference_steps,
#         guidance_scale=3.5,
#         height=control_image.size[1],
#         width=control_image.size[0],
#         generator=generator,
#     ).images[0]

#     # Resize output image back to the original dimensions if needed
#     if was_resized:
#         original_size = (input_image.width * upscale_factor, input_image.height * upscale_factor)
#         image = image.resize(original_size)

#     # Convert the output image to base64
#     buffered = io.BytesIO()
#     image.save(buffered, format="JPEG")
#     image_base64 = base64.b64encode(buffered.getvalue()).decode("utf-8")

#     logger.info("Inference completed")
#     return image_base64

# # Define Gradio interface
# def gradio_interface(input_image, upscale_factor=4, seed=42, num_inference_steps=28, controlnet_conditioning_scale=0.6):
#     if randomize_seed:
#         seed = random.randint(0, MAX_SEED)
#         logger.info("Seed randomized to: %d", seed)

#     # Run inference
#     output_image_base64 = run_inference(input_image, upscale_factor, seed, num_inference_steps, controlnet_conditioning_scale)
    
#     return Image.open(io.BytesIO(base64.b64decode(output_image_base64)))

# # Create Gradio interface
# iface = gr.Interface(
#     fn=gradio_interface,
#     inputs=[
#         gr.Image(type="pil", label="Input Image"),
#         gr.Slider(min=1, max=8, step=1, label="Upscale Factor"),
#         gr.Slider(min=0, max=MAX_SEED, step=1, label="Seed"),
#         gr.Slider(min=1, max=100, step=1, label="Inference Steps"),
#         gr.Slider(min=0.0, max=1.0, step=0.1, label="ControlNet Conditioning Scale")
#     ],
#     outputs=gr.Image(label="Output Image"),
#     title="ControlNet Image Upscaling",
#     description="Upload an image to upscale using the ControlNet model."
# )

# # Launch Gradio app
# if __name__ == '__main__':
#     iface.launch()


import logging
import random
import warnings
import gradio as gr
import os
import shutil
import spaces
import subprocess
import torch
import numpy as np
from diffusers import FluxControlNetModel
from diffusers.pipelines import FluxControlNetPipeline
from PIL import Image
from huggingface_hub import snapshot_download, login
import io
import base64
from concurrent.futures import ThreadPoolExecutor

# Configure logging
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)

# ThreadPoolExecutor for managing image processing threads
executor = ThreadPoolExecutor()

# Determine the device (GPU or CPU)
if torch.cuda.is_available():
    device = "cuda"
    logger.info("CUDA is available. Using GPU.")
else:
    device = "cpu"
    logger.info("CUDA is not available. Using CPU.")

# Load model from Huggingface Hub
huggingface_token = os.getenv("HUGGINGFACE_TOKEN")
if huggingface_token:
    login(token=huggingface_token)
    logger.info("Hugging Face token found and logged in.")
else:
    logger.warning("Hugging Face token not found in environment variables.")

# Download model using snapshot_download
model_path = snapshot_download(
    repo_id="black-forest-labs/FLUX.1-dev",
    repo_type="model",
    ignore_patterns=["*.md", "*..gitattributes"],
    local_dir="FLUX.1-dev",
    token=huggingface_token
)
logger.info("Model downloaded to: %s", model_path)

# Load pipeline
logger.info('Loading ControlNet model.')
controlnet = FluxControlNetModel.from_pretrained(
    "jasperai/Flux.1-dev-Controlnet-Upscaler", torch_dtype=torch.bfloat16
).to(device)
logger.info("ControlNet model loaded successfully.")

logger.info('Loading pipeline.')
pipe = FluxControlNetPipeline.from_pretrained(
    model_path, controlnet=controlnet, torch_dtype=torch.bfloat16
).to(device)
logger.info("Pipeline loaded successfully.")

MAX_SEED = 1000000
MAX_PIXEL_BUDGET = 1024 * 1024

@spaces.GPU
def process_input(input_image, upscale_factor):
    w, h = input_image.size
    aspect_ratio = w / h
    was_resized = False

    # Resize if input size exceeds the maximum pixel budget
    if w * h * upscale_factor**2 > MAX_PIXEL_BUDGET:
        warnings.warn(f"Requested output image is too large. Resizing to fit within pixel budget.")
        input_image = input_image.resize(
            (
                int(aspect_ratio * MAX_PIXEL_BUDGET**0.5 // upscale_factor),
                int(MAX_PIXEL_BUDGET**0.5 // aspect_ratio // upscale_factor),
            )
        )
        was_resized = True

    # Adjust dimensions to be a multiple of 8
    w, h = input_image.size
    w = w - w % 8
    h = h - h % 8

    return input_image.resize((w, h)), was_resized
    

@spaces.GPU
def run_inference(input_image, upscale_factor, seed, num_inference_steps, controlnet_conditioning_scale):
    logger.info("Processing inference.")
    input_image, was_resized = process_input(input_image, upscale_factor)

    # Rescale image for ControlNet processing
    w, h = input_image.size
    control_image = input_image.resize((w * upscale_factor, h * upscale_factor))

    # Set the random generator for inference
    generator = torch.Generator().manual_seed(seed)

    # Perform inference using the pipeline
    logger.info("Running pipeline.")
    image = pipe(
        prompt="",
        control_image=control_image,
        controlnet_conditioning_scale=controlnet_conditioning_scale,
        num_inference_steps=num_inference_steps,
        guidance_scale=3.5,
        height=control_image.size[1],
        width=control_image.size[0],
        generator=generator,
    ).images[0]

    # Resize output image back to the original dimensions if needed
    if was_resized:
        original_size = (input_image.width * upscale_factor, input_image.height * upscale_factor)
        image = image.resize(original_size)

    return image

def run_gradio_app():
    with gr.Blocks() as app:
        gr.Markdown("## Image Upscaler using ControlNet")

        # Define the inputs and outputs
        input_image = gr.Image(type="pil", label="Input Image")
        upscale_factor = gr.Slider(minimum=1, maximum=8, step=1, label="Upscale Factor")
        seed = gr.Slider(minimum=0, maximum=100, step=1, label="Seed")
        num_inference_steps = gr.Slider(minimum=1, maximum=100, step=1, label="Inference Steps")
        controlnet_conditioning_scale = gr.Slider(minimum=0.0, maximum=1.0, step=0.1, label="ControlNet Conditioning Scale")

        output_image = gr.Image(type="pil", label="Output Image")

        # Create a button to trigger the processing
        submit_button = gr.Button("Upscale Image")

        # Define the function to run when the button is clicked
        submit_button.click(run_inference, 
                             inputs=[input_image, upscale_factor, seed, num_inference_steps, controlnet_conditioning_scale], 
                             outputs=output_image)

    app.launch()

if __name__ == "__main__":
    run_gradio_app()