import gradio as gr
from openai import OpenAI

# Initialize the OpenAI client
client = OpenAI(
    api_key="EMPTY",
    base_url='https://hub.societyai.com/models/llama-3-2-3b/openai/v1',
)

with gr.Blocks(css="footer {visibility: hidden}") as demo:
    chatbot = gr.Chatbot(type="messages")
    msg = gr.Textbox()
    clear = gr.Button("Clear")

    # System message to provide initial context to the conversation
    system_message = {"role": "system", "content": "You are a helpful AI assistant expert on SEO. Please assist the user in a friendly and informative manner. As an Expert SEO Content Creator: When the user provides a keyword or phrase, your task is to create a well-optimized, high-quality SEO article for that keyword. If the user’s request is unclear, begin by asking: "What keyword are you trying to optimize for?" Detailed Steps: Clarify Keyword and Intent: After receiving the primary keyword, ask: "What is the search intent behind this keyword? (Informational, Navigational, Commercial, or Transactional)" "Are there any secondary keywords or related topics you’d like to include?" Content Structure & Planning: Propose a working title that includes the primary keyword, an outline with clear H2 and H3 headings, and an estimated word count. Seek user approval or feedback before proceeding. External Linking Strategy: Explain the importance of authoritative external links, suggest 2-3 relevant external links per 1,000 words, and ask if there are any preferred or avoided sources. Content Creation: Include the primary keyword in the first 100 words, the title, and at least one H2 heading; incorporate secondary keywords naturally; maintain a 1-2% primary keyword density; use transitional phrases, short paragraphs, bullet points; insert external links with descriptive anchor text; optimize for featured snippets. On-Page SEO Elements: Provide a meta description (150-160 characters) with the primary keyword; suggest an SEO-friendly URL; propose internal links; recommend image alt text with keywords. User Review & Revisions: Present the draft to the user, incorporate feedback, and make revisions. Final Check: Ensure all agreed-upon SEO elements are in place and confirm with the user before delivering the final version. Important Reminder: Balance SEO best practices with natural, engaging, and authoritative content to genuinely help readers, prioritizing clarity, accuracy, and user satisfaction."}

    def user(user_message, history: list):
        """Appends the user message to the conversation history."""
        if not history:
            # Initialize with system message if history is empty
            history = [system_message]
        return "", history + [{"role": "user", "content": user_message}]

    def bot(history: list):
        """Sends the conversation history to the vLLM API and streams the assistant's response."""
        # Append an empty assistant message to history to fill in as we receive the response
        history.append({"role": "assistant", "content": ""})

        try:
            # Create a chat completion with streaming enabled using the client
            completion = client.chat.completions.create(
                model="llama-3.2-3B-instruct",  # Adjust the model name if needed
                messages=history,
                stream=True
            )

            # Iterate over the streamed response
            for chunk in completion:
                # Access the delta content from the chunk
                delta = chunk.choices[0].delta
                content = getattr(delta, 'content', '')
                if content:
                    # Update the assistant's message with new content
                    history[-1]['content'] += content
                    yield history
        except Exception as e:
            # Handle exceptions and display an error message
            history[-1]['content'] += f"\n[Error]: {str(e)}"
            yield history

    # Set up the Gradio interface components
    msg.submit(user, [msg, chatbot], [msg, chatbot], queue=False).then(
        bot, chatbot, chatbot
    )
    clear.click(lambda: None, None, chatbot, queue=False)

if __name__ == "__main__":
    demo.launch()