paper_chat_tab.py

import gradio as gr
from PyPDF2 import PdfReader
from bs4 import BeautifulSoup
import openai
import traceback
import requests
from io import BytesIO
from transformers import AutoTokenizer
import json

import os
from openai import OpenAI

# Cache for tokenizers to avoid reloading
tokenizer_cache = {}

# Global variables for providers
PROVIDERS = {
    "SambaNova": {
        "name": "SambaNova",
        "logo": "https://venturebeat.com/wp-content/uploads/2020/02/SambaNovaLogo_H_F.jpg",
        "endpoint": "https://api.sambanova.ai/v1/",
        "api_key_env_var": "SAMBANOVA_API_KEY",
        "models": [
            "Meta-Llama-3.1-70B-Instruct",
            # Add more models if needed
        ],
        "type": "tuples",
        "max_total_tokens": "50000",
    },
    "Hyperbolic": {
        "name": "hyperbolic",
        "logo": "https://www.nftgators.com/wp-content/uploads/2024/07/Hyperbolic.jpg",
        "endpoint": "https://api.hyperbolic.xyz/v1",
        "api_key_env_var": "HYPERBOLIC_API_KEY",
        "models": [
            "meta-llama/Meta-Llama-3.1-405B-Instruct",
        ],
        "type": "tuples",
        "max_total_tokens": "50000",
    },
}


# Function to fetch paper information from OpenReview
def fetch_paper_info_neurips(paper_id):
    url = f"https://openreview.net/forum?id={paper_id}"
    response = requests.get(url)
    if response.status_code != 200:
        return None

    html_content = response.content
    soup = BeautifulSoup(html_content, 'html.parser')

    # Extract title
    title_tag = soup.find('h2', class_='citation_title')
    title = title_tag.get_text(strip=True) if title_tag else 'Title not found'

    # Extract authors
    authors = []
    author_div = soup.find('div', class_='forum-authors')
    if author_div:
        author_tags = author_div.find_all('a')
        authors = [tag.get_text(strip=True) for tag in author_tags]
    author_list = ', '.join(authors) if authors else 'Authors not found'

    # Extract abstract
    abstract_div = soup.find('strong', text='Abstract:')
    if abstract_div:
        abstract_paragraph = abstract_div.find_next_sibling('div')
        abstract = abstract_paragraph.get_text(strip=True) if abstract_paragraph else 'Abstract not found'
    else:
        abstract = 'Abstract not found'

    # Construct preamble in Markdown
    preamble = f"**[{title}](https://openreview.net/forum?id={paper_id})**\n\n{author_list}\n\n"

    return preamble


def fetch_paper_content_arxiv(paper_id):
    try:
        # Construct the URL for the arXiv PDF
        url = f"https://arxiv.org/pdf/{paper_id}.pdf"

        # Fetch the PDF
        response = requests.get(url)
        response.raise_for_status()  # Raise an exception for HTTP errors

        # Read the PDF content
        pdf_content = BytesIO(response.content)
        reader = PdfReader(pdf_content)

        # Extract text from the PDF
        text = ""
        for page in reader.pages:
            text += page.extract_text()

        return text  # Return full text; truncation will be handled later
    except Exception as e:
        print(f"Error fetching paper content: {e}")
        return None


def fetch_paper_content(paper_id):
    try:
        # Construct the URL
        url = f"https://openreview.net/pdf?id={paper_id}"

        # Fetch the PDF
        response = requests.get(url)
        response.raise_for_status()  # Raise an exception for HTTP errors

        # Read the PDF content
        pdf_content = BytesIO(response.content)
        reader = PdfReader(pdf_content)

        # Extract text from the PDF
        text = ""
        for page in reader.pages:
            text += page.extract_text()

        return text  # Return full text; truncation will be handled later

    except Exception as e:
        print(f"An error occurred: {e}")
        return None


def create_chat_interface(provider_dropdown, model_dropdown, paper_content, hf_token_input, default_type,
                          provider_max_total_tokens):
    # Define the function to handle the chat
    print("the type is", default_type.value)

    def get_fn(message, history, paper_content_value, hf_token_value, provider_name_value, model_name_value,
               max_total_tokens):
        provider_info = PROVIDERS[provider_name_value]
        endpoint = provider_info['endpoint']
        api_key_env_var = provider_info['api_key_env_var']
        models = provider_info['models']
        max_total_tokens = int(max_total_tokens)

        # Load tokenizer and cache it
        tokenizer_key = f"{provider_name_value}_{model_name_value}"
        if tokenizer_key not in tokenizer_cache:
            # Load the tokenizer; adjust the model path based on the provider and model
            # This is a placeholder; you need to provide the correct tokenizer path
            tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-3.2-1B-Instruct",
                                                      token=os.environ.get("HF_TOKEN"))
            tokenizer_cache[tokenizer_key] = tokenizer
        else:
            tokenizer = tokenizer_cache[tokenizer_key]

        # Include the paper content as context
        if paper_content_value:
            context = f"The discussion is about the following paper:\n{paper_content_value}\n\n"
        else:
            context = ""

        # Tokenize the context
        context_tokens = tokenizer.encode(context)
        context_token_length = len(context_tokens)

        # Prepare the messages without context
        messages = []
        message_tokens_list = []
        total_tokens = context_token_length  # Start with context tokens

        for user_msg, assistant_msg in history:
            # Tokenize user message
            user_tokens = tokenizer.encode(user_msg)
            messages.append({"role": "user", "content": user_msg})
            message_tokens_list.append(len(user_tokens))
            total_tokens += len(user_tokens)

            # Tokenize assistant message
            if assistant_msg:
                assistant_tokens = tokenizer.encode(assistant_msg)
                messages.append({"role": "assistant", "content": assistant_msg})
                message_tokens_list.append(len(assistant_tokens))
                total_tokens += len(assistant_tokens)

        # Tokenize the new user message
        message_tokens = tokenizer.encode(message)
        messages.append({"role": "user", "content": message})
        message_tokens_list.append(len(message_tokens))
        total_tokens += len(message_tokens)

        # Check if total tokens exceed the maximum allowed tokens
        if total_tokens > max_total_tokens:
            # Attempt to truncate the context first
            available_tokens = max_total_tokens - (total_tokens - context_token_length)
            if available_tokens > 0:
                # Truncate the context to fit the available tokens
                truncated_context_tokens = context_tokens[:available_tokens]
                context = tokenizer.decode(truncated_context_tokens)
                context_token_length = available_tokens
                total_tokens = total_tokens - len(context_tokens) + context_token_length
            else:
                # Not enough space for context; remove it
                context = ""
                total_tokens -= context_token_length
                context_token_length = 0

        # If total tokens still exceed the limit, truncate the message history
        while total_tokens > max_total_tokens and len(messages) > 1:
            # Remove the oldest message
            removed_message = messages.pop(0)
            removed_tokens = message_tokens_list.pop(0)
            total_tokens -= removed_tokens

        # Rebuild the final messages list including the (possibly truncated) context
        final_messages = []
        if context:
            final_messages.append(
                {"role": "system", "content": f"{context}"})
        final_messages.extend(messages)

        # Use the provider's API key
        api_key = hf_token_value or os.environ.get(api_key_env_var)
        if not api_key:
            raise ValueError("API token is not provided.")

        # Initialize the OpenAI client with the provider's endpoint
        client = OpenAI(
            base_url=endpoint,
            api_key=api_key,
        )

        try:
            # Create the chat completion
            completion = client.chat.completions.create(
                model=model_name_value,
                messages=final_messages,
                stream=True,
            )
            response_text = ""
            for chunk in completion:
                delta = chunk.choices[0].delta.content or ""
                response_text += delta
                yield response_text
        except json.JSONDecodeError as e:
            print("Failed to decode JSON during the completion creation process.")
            print(f"Error Message: {e.msg}")
            print(f"Error Position: Line {e.lineno}, Column {e.colno} (Character {e.pos})")
            print(f"Problematic JSON Data: {e.doc}")
            yield f"{e.doc}"
        except openai.OpenAIError as openai_err:
            # Handle other OpenAI-related errors
            print(f"An OpenAI error occurred: {openai_err}")
            yield f"{openai_err}"
        except Exception as ex:
            # Handle any other exceptions
            print(f"An unexpected error occurred: {ex}")
            yield f"{ex}"

    # Create the Chatbot separately to access it later
    chatbot = gr.Chatbot(
        label="Chatbot",
        scale=1,
        height=400,
        autoscroll=True,
    )

    # Create the ChatInterface
    chat_interface = gr.ChatInterface(
        fn=get_fn,
        chatbot=chatbot,
        additional_inputs=[paper_content, hf_token_input, provider_dropdown, model_dropdown, provider_max_total_tokens],
        type="tuples",
    )
    return chat_interface, chatbot


def paper_chat_tab(paper_id, paper_from):
    with gr.Column():

        # Preamble message to hint the user
        gr.Markdown("**Note:** Providing your own API token can help you avoid rate limits.")

        # Input for API token
        provider_names = list(PROVIDERS.keys())
        default_provider = provider_names[0]

        default_type = gr.State(value=PROVIDERS[default_provider]["type"])
        default_max_total_tokens = gr.State(value=PROVIDERS[default_provider]["max_total_tokens"])

        provider_dropdown = gr.Dropdown(
            label="Select Provider",
            choices=provider_names,
            value=default_provider
        )

        hf_token_input = gr.Textbox(
            label=f"Enter your {default_provider} API token (optional)",
            type="password",
            placeholder=f"Enter your {default_provider} API token to avoid rate limits"
        )

        # Dropdown for selecting the model
        model_dropdown = gr.Dropdown(
            label="Select Model",
            choices=PROVIDERS[default_provider]['models'],
            value=PROVIDERS[default_provider]['models'][0]
        )

        # Placeholder for the provider logo
        logo_html = gr.HTML(
            value=f'<img src="{PROVIDERS[default_provider]["logo"]}" width="100px" />'
        )

        # Note about the provider
        note_markdown = gr.Markdown(f"**Note:** This model is supported by {default_provider}.")

        # State to store the paper content
        paper_content = gr.State()

        # Textbox to display the paper title and authors
        content = gr.Markdown(value="")

        # Create the chat interface and get the chatbot component
        chat_interface, chatbot = create_chat_interface(provider_dropdown, model_dropdown, paper_content,
                                                        hf_token_input,
                                                        default_type, default_max_total_tokens)

        # Function to update models and logo when provider changes
        def update_provider(selected_provider):
            provider_info = PROVIDERS[selected_provider]
            models = provider_info['models']
            logo_url = provider_info['logo']
            chatbot_message_type = provider_info['type']
            max_total_tokens = provider_info['max_total_tokens']

            # Update the models dropdown
            model_dropdown_choices = gr.update(choices=models, value=models[0])

            # Update the logo image
            logo_html_content = f'<img src="{logo_url}" width="100px" />'
            logo_html_update = gr.update(value=logo_html_content)

            # Update the note markdown
            note_markdown_update = gr.update(value=f"**Note:** This model is supported by {selected_provider}.")

            # Update the hf_token_input label and placeholder
            hf_token_input_update = gr.update(
                label=f"Enter your {selected_provider} API token (optional)",
                placeholder=f"Enter your {selected_provider} API token to avoid rate limits"
            )

            # Reset the chatbot history
            chatbot_reset = []  # This resets the chatbot conversation

            return model_dropdown_choices, logo_html_update, note_markdown_update, hf_token_input_update, chatbot_message_type, max_total_tokens, chatbot_reset

        provider_dropdown.change(
            fn=update_provider,
            inputs=provider_dropdown,
            outputs=[model_dropdown, logo_html, note_markdown, hf_token_input, default_type, default_max_total_tokens,
                     chatbot],
            queue=False
        )

        # Function to update the paper info
        def update_paper_info(paper_id_value, paper_from_value, selected_model):
            if paper_from_value == "neurips":
                preamble = fetch_paper_info_neurips(paper_id_value)
                text = fetch_paper_content(paper_id_value)
                if preamble is None:
                    preamble = "Paper not found or could not retrieve paper information."
                if text is None:
                    return preamble, None, []
                return preamble, text, []
            elif paper_from_value == "paper_page":
                # Fetch the paper information from Hugging Face API
                url = f"https://huggingface.co/api/papers/{paper_id_value}?field=comments"
                response = requests.get(url)
                if response.status_code != 200:
                    return "Paper not found or could not retrieve paper information.", None, []
                paper_info = response.json()

                # Extract required information
                title = paper_info.get('title', 'No Title')
                link = f"https://huggingface.co/papers/{paper_id_value}"
                authors_list = [author.get('name', 'Unknown') for author in paper_info.get('authors', [])]
                authors = ', '.join(authors_list)
                summary = paper_info.get('summary', 'No Summary')
                num_comments = len(paper_info.get('comments', []))
                num_upvotes = paper_info.get('upvotes', 0)

                # Format the preamble
                preamble = f"🤗 [paper-page]({link})<br/>"
                preamble += f"**Title:** {title}<br/>"
                preamble += f"**Authors:** {authors}<br/>"
                preamble += f"**Summary:**<br/>>\n{summary}<br/>"
                preamble += f"👍{num_comments} 💬{num_upvotes} <br/>"

                # Fetch the paper content
                text = fetch_paper_content_arxiv(paper_id_value)
                if text is None:
                    text = "Paper content could not be retrieved."
                return preamble, text, []
            else:
                return "", "", []

        # Update paper content when paper ID changes
        paper_id.change(
            fn=update_paper_info,
            inputs=[paper_id, paper_from, model_dropdown],
            outputs=[content, paper_content, chatbot]
        )


def main():
    """
    Launches the Gradio app.
    """
    with gr.Blocks(css_paths="style.css") as demo:
        # Create an input for paper_id
        paper_id = gr.Textbox(label="Paper ID", value="")

        # Create an input for paper_from (e.g., 'neurips' or 'paper_page')
        paper_from = gr.Radio(
            label="Paper Source",
            choices=["neurips", "paper_page"],
            value="neurips"
        )

        # Build the paper chat tab
        paper_chat_tab(paper_id, paper_from)

    demo.launch(ssr_mode=False)


# Run the main function when the script is executed
if __name__ == "__main__":
    main()