app.py

from PIL import Image, ImageDraw, ImageFont
import tempfile
import gradio as gr
from smolagents import CodeAgent, InferenceClientModel, TransformersModel 
from smolagents import DuckDuckGoSearchTool, Tool
from huggingface_hub import InferenceClient
from diffusers import DiffusionPipeline
import torch

from huggingface_hub import login
import os

token = os.environ.get("HF_TOKEN")
if token:
    login(token=token)
else:
    print("Warning: HF_TOKEN not set. You may not be able to access private models or tools.")


# =========================================================
# Utility functions
# =========================================================

def add_label_to_image(image, label):
    draw = ImageDraw.Draw(image)
    font_path = "/usr/share/fonts/truetype/dejavu/DejaVuSans-Bold.ttf"
    font_size = 30
    try:
        font = ImageFont.truetype(font_path, font_size)
    except:
        font = ImageFont.load_default()

    text_bbox = draw.textbbox((0, 0), label, font=font)
    text_width, text_height = text_bbox[2] - text_bbox[0], text_bbox[3] - text_bbox[1]
    position = (image.width - text_width - 20, image.height - text_height - 20)

    rect_margin = 10
    rect_position = [
        position[0] - rect_margin,
        position[1] - rect_margin,
        position[0] + text_width + rect_margin,
        position[1] + text_height + rect_margin,
    ]
    draw.rectangle(rect_position, fill=(0, 0, 0, 128))
    draw.text(position, label, fill="white", font=font)
    return image


def plot_and_save_agent_image(agent_image, label, save_path=None):
    pil_image = agent_image.to_raw()
    labeled_image = add_label_to_image(pil_image, label)
    labeled_image.show()
    if save_path:
        labeled_image.save(save_path)
        print(f"Image saved to {save_path}")
    else:
        print("No save path provided. Image not saved.")


def generate_prompts_for_object(object_name):
    return {
        "past": f"Show an old version of a {object_name} from its early days.",
        "present": f"Show a {object_name} with current features/design/technology.",
        "future": f"Show a futuristic version of a {object_name}, by predicting advanced features and futuristic design."
    }

# =========================================================
# Tool wrapper for m-ric/text-to-image
# =========================================================
'''
class WrappedTextToImageTool(Tool):
    name = "text_to_image"
    description = "Generates an image from a text prompt using the m-ric/text-to-image tool."
    inputs = {
        "prompt": {
            "type": "string",
            "description": "Text prompt to generate an image"
        }
    }
    output_type = "image"

    def __init__(self):
        self.client = InferenceClient("m-ric/text-to-image")

    def forward(self, prompt):
        return self.client.text_to_image(prompt)

'''

'''
class TextToImageTool(Tool):
    description = "This tool creates an image according to a prompt, which is a text description."
    name = "image_generator"
    inputs = {"prompt": {"type": "string", "description": "The image generator prompt. Don't hesitate to add details in the prompt to make the image look better, like 'high-res, photorealistic', etc."}}
    output_type = "image"
    model_sdxl = "black-forest-labs/FLUX.1-schnell"
    client = InferenceClient(model_sdxl, provider="replicate")


    def forward(self, prompt):
        return self.client.text_to_image(prompt)
'''
'''
class TextToImageTool(Tool):
    description = "This tool creates an image according to a prompt. Add details like 'high-res, photorealistic'."
    name = "image_generator"
    inputs = {
        "prompt": {
            "type": "string",
            "description": "The image generation prompt"
        }
    }
    output_type = "image"

    def __init__(self):
        super().__init__()
        dtype = torch.bfloat16
        device = "cuda" if torch.cuda.is_available() else "cpu"
        print(f"Using device: {device}")
        self.pipe = DiffusionPipeline.from_pretrained(
            "aiyouthalliance/Free-Image-Generation-CC0", 
            torch_dtype=dtype
        ).to(device)

    def forward(self, prompt):
        image = self.pipe(prompt).images[0]
        return image
'''
image_generation_tool = Tool.from_space(
    "black-forest-labs/FLUX.1-schnell",
    api_name="/infer",  # Optional if there's only one endpoint
    name="image_generator",
    description="Generate an image from a prompt"
)


# =========================================================
# Tool and Agent Initialization
# =========================================================
#image_generation_tool= TextToImageTool()
#image_generation_tool = WrappedTextToImageTool()
search_tool = DuckDuckGoSearchTool()
#print('iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii')
#llm_engine = InferenceClientModel("Qwen/Qwen2.5-72B-Instruct")
#llm_engine = TransformersModel(
#    model_id="Qwen/Qwen2.5-72B-Instruct",
#    device="cuda",
#    max_new_tokens=5000,
#)

#from smolagents import LiteLLMModel

#llm_engine  = LiteLLMModel(model_id="Qwen/Qwen2.5-72B-Instruct", temperature=0.2, max_tokens=5000)
#llm_engine=InferenceClientModel()

llm_engine = InferenceClientModel("Qwen/Qwen2.5-Coder-32B-Instruct")
agent = CodeAgent(tools=[image_generation_tool, search_tool], model=llm_engine)

# =========================================================
# Main logic for image generation
# =========================================================

def generate_object_history(object_name):
    images = []
    prompts = generate_prompts_for_object(object_name)
    labels = {
        "past": f"{object_name} - Past",
        "present": f"{object_name} - Present",
        "future": f"{object_name} - Future"
    }


    general_instruction = (
        "Search the necessary information and features for the following prompt, "
        "then generate an image of it."
    )

    for time_period, prompt in prompts.items():
        print(f"Generating {time_period} frame: {prompt}")
        #result = agent.run(prompt)

        try:
            result = agent.run(
                general_instruction,
                additional_args={"user_prompt": prompt}
            )
            image = result.to_raw()
        except Exception as e:
            print(f"Agent failed on {time_period}: {e}")
            continue
            
        images.append(result.to_raw())
        image_filename = f"{object_name}_{time_period}.png"
        plot_and_save_agent_image(result, labels[time_period], save_path=image_filename)

    gif_path = f"{object_name}_evolution.gif"
    images[0].save(gif_path, save_all=True, append_images=images[1:], duration=1000, loop=0)
    return [(f"{object_name}_past.png", labels["past"]),
            (f"{object_name}_present.png", labels["present"]),
            (f"{object_name}_future.png", labels["future"])], gif_path
    #return images, gif_path

# =========================================================
# Gradio Interface
# =========================================================

def create_gradio_interface():
    with gr.Blocks() as demo:
        gr.Markdown("# TimeMetamorphy: An Object Evolution Generator")
        gr.Markdown("""
        Explore how everyday objects evolved over time. Enter an object name like "phone", "car", or "bicycle"
        and see its past, present, and future visualized with AI!
        """)

        default_images = [
            ("car_past.png", "Car - Past"),
            ("car_present.png", "Car - Present"),
            ("car_future.png", "Car - Future")
        ]
        default_gif_path = "car_evolution.gif"

        with gr.Row():
            with gr.Column():
                object_name_input = gr.Textbox(label="Enter an object name", placeholder="e.g. bicycle, car, phone")
                generate_button = gr.Button("Generate Evolution")
                image_gallery = gr.Gallery(label="Generated Images", columns=3, rows=1, value=default_images)
                #image_gallery = gr.Gallery(label="Generated Images", columns=3, rows=1, type="filepath")
                gif_output = gr.Image(label="Generated GIF", value=default_gif_path)

        generate_button.click(fn=generate_object_history, inputs=[object_name_input], outputs=[image_gallery, gif_output])

    return demo

# =========================================================
# Run the app
# =========================================================

demo = create_gradio_interface()
demo.launch(share=True)