app.py

import gradio as gr
import json
from transformers import pipeline

# Load a text generation model from Hugging Face without an API key
# You can use a smaller model if gpt-3.5-turbo isn't available
generator = pipeline("text-generation", model="gpt-3.5-turbo")

# Function to attempt to fix JSON using a Hugging Face model
def ai_fix_json(json_data):
    # Prepare a prompt for the AI model to fix the JSON
    prompt = f"Fix the following JSON data and make it valid:\n\n{json_data}\n\nFixed JSON:"

    # Generate a response from the model
    response = generator(prompt, max_length=1024, num_return_sequences=1)[0]['generated_text']

    # Extract the fixed JSON from the response (you may need to fine-tune this depending on the model's output format)
    fixed_json = response.split("Fixed JSON:")[-1].strip()
    
    # Try to load the fixed JSON to ensure it's valid
    try:
        parsed_data = json.loads(fixed_json)
        pretty_json = json.dumps(parsed_data, indent=4)
        return pretty_json, "JSON fixed using AI."
    
    except json.JSONDecodeError as e:
        return None, f"Failed to fix JSON: {str(e)}"

# Function to handle file uploads and processing
def process_file(uploaded_file):
    # Read the content of the uploaded file directly
    json_data = uploaded_file  # This is already the content of the file as a string
    
    # Attempt to fix the JSON using AI
    cleaned_json, message = ai_fix_json(json_data)
    
    if cleaned_json:
        # Return the fixed JSON for display and as a downloadable file
        return cleaned_json, message, cleaned_json
    
    else:
        return None, message, None

# Gradio interface
iface = gr.Interface(
    fn=process_file,
    inputs=gr.File(label="Upload your JSON file"),
    outputs=[gr.JSON(label="Fixed JSON"), "text", gr.File(label="Download cleaned JSON file")],
    title="AI-Powered JSON Cleaner",
    description="Upload a JSON file to automatically fix, remove duplicates, and download the cleaned version using AI."
)

iface.launch()