upscalecreateimages.py

import os
import openai
import pandas as pd
from tqdm import tqdm
from PIL import Image
import requests
import io

# Set OpenAI and Stability.ai API keys
openai.api_key = 'YOUR_OPEN_AI_KEY'
stability_ai_key = 'YOUR_STABILITY_AI_KEY'

# Set Stability API key for image upscaling
api_key = 'YOUR_STABILIT_AI_KEY'
if api_key is None:
    raise Exception("Missing Stability API key.")
api_host = os.getenv("API_HOST", "https://api.stability.ai")


def generate_clickable_title(detail):
    prompt = f"Generate a catchy and clickable title for a Cork Back Coaster with the theme: '{detail}'. Maximum 50 characters. At the end of each title write Cork Back Coaster"
    response = openai.ChatCompletion.create(
        model="gpt-3.5-turbo",
        messages=[
            {"role": "system", "content": "You are a helpful assistant."},
            {"role": "user", "content": prompt}
        ]
    )
    clickable_title = response['choices'][0]['message']['content'].strip()
    clickable_title = clickable_title.replace('"', '')  # Remove double quotes
    return clickable_title

def generate_description(detail):
    prompt = f"Generate a compelling description for a Cork Back Coaster with the theme: '{detail}'. Maximum 150 characters."
    response = openai.ChatCompletion.create(
        model="gpt-3.5-turbo",
        messages=[
            {"role": "system", "content": "You are a helpful assistant."},
            {"role": "user", "content": prompt}
        ]
    )
    description = response['choices'][0]['message']['content'].strip()
    description = description.replace('"', '')  # Remove double quotes
    description += """
    <p>This personalized photo coaster helps you market yourself the right way or uplift any table with a custom touch. Each round and square coaster is made with a high-gloss, polyester-coated hardboard top, and features a genuine cork bottom to prevent sliding. </p>
<p>.: Material: Genuine cork bottom finished with a glossy white top made of polyester-coated hardboard<br />.: Size: 3.75&Prime; x 3.75&Prime; (Square) and 4&Prime; x 4&Prime; (Round)<br />.: High-gloss top<br />.: NB! One coaster per listing<br />.: Assembled in the USA from globally sourced parts</p>
    """
    return description

def generate_tags(detail):
    prompt = f"Generate relevant tags for a Cork Back Coaster with the theme: '{detail}'. Separate the tags with commas."
    response = openai.ChatCompletion.create(
        model="gpt-3.5-turbo",
        messages=[
            {"role": "system", "content": "You are a helpful assistant."},
            {"role": "user", "content": prompt}
        ]
    )
    tag = response['choices'][0]['message']['content'].strip()
    tag = tag.replace('"', '')  # Remove double quotes
    return tag

csv_path = "input.csv"  
df = pd.read_csv(csv_path)

file_names = []
local_paths = []
titles = []
descriptions = []
tags = []

for idx, row in tqdm(df.iterrows(), total=df.shape[0]):
    detail = row['details']
    title = generate_clickable_title(detail)
    description = generate_description(detail)
    tag = generate_tags(detail)

    # Use the detail from the CSV as the image prompt
    image_prompt = detail

    # Generate the image using stable diffusion
    
    url = "https://api.stability.ai/v1/generation/stable-diffusion-v1-5/text-to-image"
    headers = {"Authorization": f"Bearer {stability_ai_key}", "Accept": "image/png"}
    data = {
        "width": 512,
        "height": 512,
        "text_prompts": [
            {
                "text": image_prompt,
                "weight": 0.5,
            
            }
        ]
    }
    response = requests.post(url, headers=headers, json=data)
    image_data = response.content
    image = Image.open(io.BytesIO(image_data))
    file_name = f"image_{idx}.png"  
    local_path = f"{file_name}"  
    image.save(local_path)

    # Upscale the image using the "esrgan-v1-x2plus" engine
    response = requests.post(
        f"{api_host}/v1/generation/esrgan-v1-x2plus/image-to-image/upscale",
        headers={
            "Accept": "image/png",
            "Authorization": f"Bearer {api_key}"
        },
        files={
            "image": open(local_path, "rb")
        },
        data={
            "width": 2048,
        }
    )

    if response.status_code != 200:
        raise Exception("Non-200 response: " + str(response.text))

    upscaled_path = f"upscaled_{file_name}"
    with open(upscaled_path, "wb") as f:
        f.write(response.content)
    local_paths.append(upscaled_path)

    file_names.append(file_name)
    titles.append(title)
    descriptions.append(description)
    tags.append(tag)

output_df = pd.DataFrame({
    "file_name": file_names,
    "local_path": local_paths,
    "title": titles,
    "description": descriptions,
    "tags": tags
})

output_df.to_csv("product_information.csv", index=False)