app.py

import gradio as gr
import base64
import requests
import re
from langchain_groq import ChatGroq

# Define API Key and Model Name
neura_code = "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"
neura_chat = "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"
MODEL_NAME = "deepseek-r1-distill-llama-70b"
CHATBOT_MODEL = "llama-3.3-70b-versatile"

groq_model = ChatGroq(temperature=0.4, groq_api_key=neura_code, model_name=MODEL_NAME)
chatbot_model = ChatGroq(temperature=0.6, groq_api_key=neura_chat, model_name=CHATBOT_MODEL)

def clean_output(text):
    """
    Removes <think>...</think> sections from the generated output.
    """
    return re.sub(r"<think>.*?</think>", "", text, flags=re.DOTALL).strip()

def ocr_gradio(image):
    """
    Processes the uploaded image, extracts Markdown content, and generates code from it.
    """
    api_key = "44b3dc5fd85497b6b0885c0cdc071b1d5804eeb9f32e03917f0d949c4a779a37"
    model = "meta-llama/Llama-3.2-90B-Vision-Instruct-Turbo"

    system_prompt = (
        "Convert the provided image into Markdown format. Include all elements like arrows, headers, footers, subtexts, and tables."
    )

    image_data = encode_image(image)
    response = requests.post(
        "https://api.together.xyz/v1/chat/completions",
        headers={"Authorization": f"Bearer {api_key}", "Content-Type": "application/json"},
        json={
            "model": model,
            "messages": [
                {"role": "user", "content": [
                    {"type": "text", "text": system_prompt},
                    {"type": "image_url", "image_url": {"url": image_data}},
                ]}
            ],
        },
    )

    if response.status_code == 200:
        markdown_text = response.json().get("choices", [{}])[0].get("message", {}).get("content", "")
        return generate_code(markdown_text)
    else:
        return f"Error: {response.status_code} - {response.text}"

def encode_image(image_path: str) -> str:
    """Encodes a local image as a base64 string."""
    with open(image_path, "rb") as image_file:
        encoded_string = base64.b64encode(image_file.read()).decode("utf-8")
    return f"data:image/jpeg;base64,{encoded_string}"

def generate_code(user_input):
    """
    Sends the extracted Markdown to LLM for code generation.
    """
    try:
        prompt = f"Convert the following Markdown into executable code.\n\n{user_input}"
        completion = groq_model.invoke(prompt)
        return clean_output(completion.content)
    except Exception as e:
        return f"Error processing input: {str(e)}"

def chatbot_response(user_message):
    """Handles chatbot interactions using the Groq model."""
    try:
        return chatbot_model.invoke(user_message).content
    except Exception as e:
        return f"Chatbot Error: {str(e)}"

custom_css = """
body { background-color: #121212; color: #E0E0E0; font-family: 'Poppins', sans-serif; }
.gradio-container { max-width: 800px; margin: auto; }
input, textarea { background: #222; color: #FFF; border-radius: 8px; }
button { background: #6200EE; color: white; font-weight: bold; border-radius: 8px; padding: 10px; }
button:hover { background: #3700B3; }
"""

gradio_theme = gr.themes.Base(
    primary_hue="purple",
    secondary_hue="indigo",
    font=["Poppins", "sans-serif"]
)

ocr_interface = gr.Blocks()

with ocr_interface:
    gr.Markdown("<h1 style='text-align: center;'>Neura.AI</h1>")
    gr.Interface(
        fn=ocr_gradio,
        inputs=gr.Image(type="filepath"),
        outputs="text",
        description="<p style='color:#BBB;'>The future of coding is here. Just upload a diagram of the technical architecture of your project or app to Neura.AI and generate high-quality, executable code!</p>",
        theme=gradio_theme,
        css=custom_css
    )

chatbot_interface = gr.Blocks()

with chatbot_interface:
    gr.Markdown("<h1 style='text-align: center;'>NeuraGPT</h1>")
    gr.Interface(
        fn=chatbot_response,
        inputs=gr.Textbox(placeholder="Ask me anything about coding!", lines=2),
        outputs="text",
        description="<p style='color:#BBB;'>Engage with NeuraGPT for AI-powered insights and smart coding advice.</p>",
        theme=gradio_theme,
        css=custom_css
    )

demo = gr.TabbedInterface([ocr_interface, chatbot_interface], ["Neura.AI", "NeuraGPT"], theme=gradio_theme)

demo.launch()