app.py

from huggingface_hub import InferenceClient
from langchain_community.llms import HuggingFaceHub
from langchain_community.tools import DuckDuckGoSearchResults
from langchain.agents import create_react_agent, AgentExecutor
from langchain_core.tools import BaseTool
from pydantic import Field
from PIL import Image, ImageDraw, ImageFont
import tempfile
import gradio as gr
from io import BytesIO
from typing import Optional


# === Image generation tool ===
class TextToImageTool(BaseTool):
    name: str = "text_to_image"
    description: str = "Generate an image from a text prompt."
    client: InferenceClient = Field(exclude=True)

    def _run(self, prompt: str) -> Image.Image:
        print(f"[Tool] Generating image for prompt: {prompt}")
        image_bytes = self.client.text_to_image(prompt)
        return Image.open(BytesIO(image_bytes))

    def _arun(self, prompt: str):
        raise NotImplementedError("This tool does not support async.")


# === Labeling Function ===
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_width, text_height = draw.textsize(label, font=font)
    position = (image.width - text_width - 20, image.height - text_height - 20)
    rect_position = [position[0] - 10, position[1] - 10, position[0] + text_width + 10, position[1] + text_height + 10]
    draw.rectangle(rect_position, fill=(0, 0, 0, 128))
    draw.text(position, label, fill="white", font=font)
    return image


# === Prompt Generator ===
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}, predicting future features/designs.",
    }


# === Agent Setup ===
text_to_image_client = InferenceClient("m-ric/text-to-image")

text_to_image_tool = TextToImageTool(client=text_to_image_client)

search_tool = DuckDuckGoSearchResults()

llm = HuggingFaceHub(
    repo_id="Qwen/Qwen2.5-72B-Instruct",
    model_kwargs={"temperature": 0.7, "max_new_tokens": 512},
)

agent = create_react_agent(llm=llm, tools=[text_to_image_tool, search_tool])
agent_executor = AgentExecutor(agent=agent, tools=[text_to_image_tool, search_tool], verbose=True)


# === History Generator ===
def generate_object_history(object_name: str):
    prompts = generate_prompts_for_object(object_name)
    images = []
    labels = {
        "past": f"{object_name} - Past",
        "present": f"{object_name} - Present",
        "future": f"{object_name} - Future"
    }
    for period, prompt in prompts.items():
        result = text_to_image_tool._run(prompt)
        labeled = add_label_to_image(result, labels[period])
        file_path = f"{object_name}_{period}.png"
        labeled.save(file_path)
        images.append((file_path, labels[period]))
    gif_path = f"{object_name}_evolution.gif"
    pil_images = [Image.open(img[0]) for img in images]
    pil_images[0].save(gif_path, save_all=True, append_images=pil_images[1:], duration=1000, loop=0)
    return images, gif_path


# === Gradio UI ===
def create_gradio_interface():
    with gr.Blocks() as demo:
        gr.Markdown("# TimeMetamorphy: Evolution Visualizer")

        with gr.Row():
            with gr.Column():
                object_input = gr.Textbox(label="Enter Object (e.g., car, phone)")
                generate_button = gr.Button("Generate Evolution")
                gallery = gr.Gallery(label="Generated Images").style(grid=3)
                gif_display = gr.Image(label="Generated GIF")

        generate_button.click(fn=generate_object_history, inputs=object_input, outputs=[gallery, gif_display])

    return demo


# === Launch App ===
demo = create_gradio_interface()
demo.launch(share=True)