app3.py

import gradio as gr
import requests
import time
import requests
import base64
import time



token = '5UAYO8UWHNQKT3UUS9H8V360L76MD72DRIUY9QC2'


    

##############################################################
#################################################

def SD_call(image_prompt, age, color, hair_color,NSFW):


    postive = "clothes"
    negative = "naked, nsfw, porn"
    serverless_api_id = '3g77weiulabzuk'
    # Define the URL you want to send the request to
    url = f"https://api.runpod.ai/v2/{serverless_api_id}/runsync"

    # Define your custom headers
    headers = {
        "Authorization": f"Bearer {token}",
        "Accept": "application/json",
        "Content-Type": "application/json"
    }
    
    # Define your data (this could also be a JSON payload)
    print("SD_processing")
    if NSFW:
        postive = "naked, nsfw"
        negative = "clothes"
    data = {
        "input": {
            "api": {
                "method": "POST",
                "endpoint": "/sdapi/v1/txt2img"
            },
            "payload": {
                "override_settings": {
                    "sd_model_checkpoint": "CyberRealistic",
                    "sd_vae": ""
                },
                "override_settings_restore_afterwards": True,
                "refiner_checkpoint": "",
                "refiner_switch_at": 0.8,
                "prompt": f"masterpiece, best quality, 8k, (looking at viewer:1.1), gorgeous, hot, seductive, {age} years old american {color} woman, (eye contact:1.1), beautiful face, hyper detailed, best quality, ultra high res, {hair_color} hair,blue eyes, photorealistic, high resolution, detailed, raw photo, 1girl,{image_prompt}, {positive} ",
                "negative_prompt": f"EasyNegative, fat, paintings, sketches, (worst quality:2), (low quality:2), (normal quality:2), lowres, ((monochrome)), ((grayscale)), bad anatomy, text, error, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry, bad feet, poorly drawn face, bad proportions, gross proportions, ng_deepnegative_v1_75t, badhandsv5-neg, {negative}",
                "seed": -1,
                "batch_size": 1,
                "steps": 30,
                "cfg_scale": 7,
                "width": 520,
                "height": 520,
                "sampler_name": "DPM++ SDE Karras",
                "sampler_index": "DPM++ SDE Karras",
                "restore_faces": False
            }
        }
    }




    # Send the POST request with headers and data
    response = requests.post(url, headers=headers, json=data)

    # Check the response
    if response.status_code == 200:
        response_data = response.json()
        msg_id = response_data['id']
        print("Message ID:", msg_id)

        # Poll the status until it's not 'IN_QUEUE'
        while response_data['status'] == 'IN_QUEUE':
            time.sleep(5)  # Wait for 5 seconds before checking again
            response = requests.get(f"{url}/{msg_id}", headers=headers)
            
            try:
                response_data = response.json()
            except Exception as e:
                print("Error decoding JSON:", e)
                print("Response content:", response.text)
                break  # Exit the loop on JSON decoding error

        # Check if the response contains images
        if 'images' in response_data.get('output', {}):
            base64_image = response_data['output']['images'][0]
            image_bytes = base64.b64decode(base64_image)
            
            # Save the image to a file
            image_path = f"output_image_{msg_id}.png"
            with open(image_path, "wb") as img_file:
                img_file.write(image_bytes)

            print(f"Image downloaded successfully: {image_path}")
            
            return image_path

        else:
            return "No images found in the response."
            
    else:
        # Print error message
        return f"Error: {response.status_code} - {response.text}"
                        



##############################################################
#################################################



def LLM_call(message_log, temperature):

    serverless_api_id = '4whzcbwuriohqh'
    # Define the URL you want to send the request to
    url = f"https://api.runpod.ai/v2/{serverless_api_id}/run"

    # Define your custom headers
    headers = {
        "Authorization": f"Bearer {token}",
        "Accept": "application/json",
        "Content-Type": "application/json"
    }

       

    # Define your data (this could also be a JSON payload)
    data = {

        "input": {
                "prompt": message_log,
                "max_new_tokens": 4500,
                "temperature": 0.7,
                "top_k": 50,
                "top_p": 0.9,
                "repetition_penalty": 1.2,
                "batch_size": 8,
                "stop": ["</s>"]
            }
    }
   
    


    # Send the POST request with headers and data
    call = requests.post(url, headers=headers, json=data)
    response_data = call.json()
    msg_id = response_data['id']
    print("Message ID:", msg_id)
    output = "Output not available"
    # Poll the API until the response is ready
    while True:
        # Get the status using the message ID
        response = requests.get(f"https://api.runpod.ai/v2/{serverless_api_id}/status/{msg_id}", headers=headers)

        if response.status_code == 200:
            response_data = response.json()
            status = response_data.get('status')

            if status == 'COMPLETED':
                # Access the 'output' directly from the response
                output = response_data.get('output', 'Output not available')
                print("Response content:", output)
                break  # Exit the loop once the response is ready
            elif status == 'FAILED':
                error_message = response_data.get('error', 'Unknown error')
                print("Request failed. Reason:", error_message)
                break  # Exit the loop if the request failed
        else:
            print("Failed to get status. HTTP status code:", response.status_code)

        # Wait for a short time before polling again (e.g., 2 seconds)
        time.sleep(2)
    
    return output





def Chat_call(chat,temperature):
    message_log = [{"role": "assistant", "content": f"I want you to act as a woman named Julie. Use colloquial language. Julie is 25 years old, she is very hot physically, she loves watching football, she loves to talk about sex. Julie loves sex and being fucked. She never refuses anything. You'll answer to a man, call him baby. Write very short answers, 15 words."},]      
    new_message = {"role": "user", "content": chat}
    message_log.append(new_message)
    response = LLM_call(message_log, temperature)
    print("From Runpod:" + response)
    return response




########################################################
#######################################################



def greet(chat, image_prompt, temperature, age, color, hair_color,NSFW):
    if image_prompt.strip():
        image_path = SD_call(image_prompt,age, color, hair_color,NSFW)
        return "Image generated successfully", image_path

    if temperature > 3:
        return "You are too warm please try again", None
    else:
        text_answer = Chat_call(chat,temperature)
        return text_answer, None

demo = gr.Interface(
    fn=greet,
    inputs=[
        "text",
        gr.Textbox(label="Image", lines=3),
        gr.Slider(label="Text temperature", value=1, minimum=0, maximum=2),
        gr.Slider(label="Age", value=22, minimum=18, maximum=75),
        gr.Dropdown(["asian", "white", "black", "latina"], label="Color", info="Will add more later!"),
        gr.Dropdown(["blond", "brune", "red", "white", "pink", "black", "blue", "green"], label="Hair color", info="Blond is cool"),
        gr.Checkbox(label="NSFW", info="👀👀👀")
    ],
    flagging_options=["blurry", "incorrect", "other"],
    outputs=[gr.Textbox(label="Answer", lines=3), gr.Image(label="Generated Image", type="filepath")],
)

demo.launch(share=True)