app.py

import gradio as gr
import os
import re
#from tempfile import NamedTemporaryFile
import numpy as np
import spaces
import random
import string
from diffusers import StableDiffusion3Pipeline
import torch
from pathos.multiprocessing import ProcessingPool as ProcessPoolExecutor
import requests
from lxml.html import fromstring

pool = ProcessPoolExecutor(4)
pool.__enter__()

#model_id = "runwayml/stable-diffusion-v1-5"
model_id = "stabilityai/stable-diffusion-3-medium-diffusers"

device = "cuda" if torch.cuda.is_available() else "cpu"
if torch.cuda.is_available():
    torch.cuda.max_memory_allocated(device=device)
    pipe = StableDiffusion3Pipeline.from_pretrained(model_id, torch_dtype=torch.float16, variant="fp16", use_safetensors=True, token=os.getenv('hf_token'))
    pipe = pipe.to(device)
else: 
    pipe = StableDiffusion3Pipeline.from_pretrained(model_id, use_safetensors=True, token=os.getenv('hf_token'))
    pipe = pipe.to(device)

def translate(text,lang):
    text = re.sub(f'[{string.punctuation}]', '', re.sub('[\s+]', ' ', text)).lower().strip()
    lang = re.sub(f'[{string.punctuation}]', '', re.sub('[\s+]', ' ', lang)).lower().strip()
    user_agents = [
        'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/126.0.0.0 Safari/537.36',
        'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/126.0.0.0 Safari/537.36',
        'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/126.0.0.0 Safari/537.36',
        'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/126.0.0.0 Safari/537.36',
        'Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/126.0.0.0 Safari/537.36',
        'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/16.1 Safari/605.1.15',
        'Mozilla/5.0 (Macintosh; Intel Mac OS X 13_1) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/16.1 Safari/605.1.15'
    ]
    url = f'https://www.google.com/search?q=translate to {lang}: {text}'
    print(url)
    resp = requests.get(
        url = url,
        headers = {
            'User-Agent': random.choice(user_agents)
        }
    )
    print(resp)
    content = resp.content
    html = fromstring(content)
    rslt = html.xpath('//pre[@aria-label="Translated text"]/span')
    translated = text
    try:
        t = rslt[0].text.strip()
        translated = t
    except:
        print(f'"{text}" is already in {lang}!')
    ret = re.sub(f'[{string.punctuation}]', '', re.sub('[\s+]', ' ', translated)).lower().strip()
    print(ret)
    return ret

def generate_random_string(length):
    characters = string.ascii_letters + string.digits
    return ''.join(random.choice(characters) for _ in range(length))

@spaces.GPU(duration=120)
def Piper(_do, _dont):
    return pipe(
        _do,
        height=480,
        width=480,
        negative_prompt=_dont,
        num_inference_steps=400,
        guidance_scale=10
    )

def infer(prompt1,prompt2,prompt3,prompt4):
    name = generate_random_string(12)+".png"
    if prompt1 == None:
        prompt1 = "any"
    else:
        prompt1 = " and ".join([translate(v,"english").upper() for v in prompt1])
    
    if prompt2 == None:
        prompt2 = "any"
    else:
        prompt2 = " and ".join([translate(v,"english").upper() for v in prompt2])
    
    if prompt3 == None:
        prompt3 = "any"
    else:
        prompt3 = " and ".join([translate(v,"english").upper() for v in prompt3])
    
    if prompt4 == None:
        prompt4 = "none"
    else:
        prompt4 = " and ".join([translate(v,"english").upper() for v in prompt4])
        
    _do = f'Show an authentic {prompt3} scene, while focusing on the details, of {prompt1}, as the main elements, and, showing {prompt2} in the background.'
    _dont = f'ANY usage of {prompt4}...'
    print(_do)
    print(_dont)
    image = Piper(_do, _dont).images[0].save(name)
    return name

css="""
#col-container {
    margin: 0 auto;
    max-width: 13cm;
}
#image-container {
    aspect-ratio: 1 / 1;
}
"""

if torch.cuda.is_available():
    power_device = "GPU"
else:
    power_device = "CPU"

with gr.Blocks(theme=gr.themes.Soft(),css=css) as demo:
    with gr.Column(elem_id="col-container"):
        gr.Markdown(f"""
            # Image Generator
            Currently running on {power_device}.
        """)
        with gr.Column():
            with gr.Row():
                prompt1 = gr.Dropdown(
                    multiselect=True,
                    allow_custom_value=True,
                    max_choices=3,
                    label="Foreground Elements",
                    show_label=True,
                    container=True
                )
            with gr.Row():
                prompt2 = gr.Dropdown(
                    multiselect=True,
                    allow_custom_value=True,
                    max_choices=4,
                    label="Background Elements",
                    show_label=True,
                    container=True
                )
            with gr.Row():
                prompt3 = gr.Dropdown(
                    multiselect=True,
                    allow_custom_value=True,
                    max_choices=2,
                    label="Background Events",
                    show_label=True,
                    container=True
                )
            with gr.Row():
                prompt4 = gr.Dropdown(
                    multiselect=True,
                    allow_custom_value=True,
                    max_choices=5,
                    label="Forbidden Elements/Events",
                    show_label=True,
                    container=True
                )
            with gr.Row():
                run_button = gr.Button("Run")
        result = gr.Image(elem_id="image-container", label="Result", show_label=False, type='filepath')
    run_button.click(
        fn = infer,
        inputs = [prompt1,prompt2,prompt3,prompt4],
        outputs = [result]
    )

demo.queue().launch()