app.py

import gradio as gr
import requests
from datetime import datetime, timezone, timedelta

API_URL = "https://huggingface.co/api/daily_papers"

class PaperManager:
    def __init__(self, papers_per_page=30):
        self.papers_per_page = papers_per_page
        self.current_page = 1
        self.papers = []
        self.total_pages = 1
        self.time_filter = 'All Time'  # Default filter

    def calculate_score(self, paper):
        """
        Calculate the score of a paper based on upvotes and age.
        This mimics the "hotness" algorithm used by platforms like Hacker News.
        """
        upvotes = paper.get('paper', {}).get('upvotes', 0)
        published_at_str = paper.get('publishedAt', datetime.now(timezone.utc).isoformat())
        try:
            published_time = datetime.fromisoformat(published_at_str.replace('Z', '+00:00'))
        except ValueError:
            # If parsing fails, use current time to minimize the impact on sorting
            published_time = datetime.now(timezone.utc)
        
        time_diff = datetime.now(timezone.utc) - published_time
        time_diff_hours = time_diff.total_seconds() / 3600  # Convert time difference to hours

        # Avoid division by zero and apply the hotness formula
        score = upvotes / ((time_diff_hours + 2) ** 1.5)
        return score

    def fetch_papers(self, time_filter='All Time'):
        """
        Fetch papers from the API and apply time filtering.
        """
        try:
            response = requests.get(f"{API_URL}?limit=100")
            response.raise_for_status()
            data = response.json()

            # Apply time filter
            filtered_data = self.apply_time_filter(data, time_filter)

            # Sort papers by calculated score descending
            self.papers = sorted(
                filtered_data,
                key=lambda x: self.calculate_score(x),
                reverse=True
            )

            self.total_pages = max((len(self.papers) + self.papers_per_page - 1) // self.papers_per_page, 1)
            self.current_page = 1
            self.time_filter = time_filter
            return True
        except requests.RequestException as e:
            print(f"Error fetching papers: {e}")
            return False
        except Exception as e:
            print(f"Unexpected error: {e}")
            return False

    def apply_time_filter(self, data, time_filter):
        """
        Filter papers based on the selected timeframe.
        """
        if time_filter == 'All Time':
            return data
        
        now = datetime.now(timezone.utc)
        if time_filter == 'Last Week':
            threshold = now - timedelta(weeks=1)
        elif time_filter == 'Last Month':
            threshold = now - timedelta(days=30)
        elif time_filter == 'Last Year':
            threshold = now - timedelta(days=365)
        else:
            # If an unknown filter is provided, default to all time
            return data

        filtered = []
        for paper in data:
            published_at_str = paper.get('publishedAt', '')
            try:
                published_time = datetime.fromisoformat(published_at_str.replace('Z', '+00:00'))
                if published_time >= threshold:
                    filtered.append(paper)
            except ValueError:
                # Skip papers with invalid date formats
                continue
        return filtered

    def format_paper(self, paper, rank):
        title = paper.get('title', 'No title')
        paper_id = paper.get('paper', {}).get('id', '')
        url = f"https://huggingface.co/papers/{paper_id}"
        authors = ', '.join([author.get('name', '') for author in paper.get('paper', {}).get('authors', [])]) or 'Unknown'
        upvotes = paper.get('paper', {}).get('upvotes', 0)
        comments = paper.get('numComments', 0)
        published_time = datetime.fromisoformat(
            paper.get('publishedAt', datetime.now(timezone.utc).isoformat()).replace('Z', '+00:00')
        )
        time_diff = datetime.now(timezone.utc) - published_time
        time_ago_days = time_diff.days
        time_ago = f"{time_ago_days} days ago" if time_ago_days > 0 else "today"

        return f"""
        <tr class="athing">
            <td align="right" valign="top" class="title"><span class="rank">{rank}.</span></td>
            <td valign="top" class="title">
                <a href="{url}" class="storylink" target="_blank">{title}</a>
            </td>
        </tr>
        <tr>
            <td colspan="1"></td>
            <td class="subtext">
                <span class="score">{upvotes} upvotes</span><br>
                authors: {authors} | {time_ago} | <a href="#">{comments} comments</a>
            </td>
        </tr>
        <tr style="height:5px"></tr>
        """

    def render_papers(self):
        start = (self.current_page - 1) * self.papers_per_page
        end = start + self.papers_per_page
        current_papers = self.papers[start:end]

        if not current_papers:
            return "<div class='no-papers'>No papers available for this page.</div>"

        papers_html = "".join([self.format_paper(paper, idx + start + 1) for idx, paper in enumerate(current_papers)])
        return f"""
        <table border="0" cellpadding="0" cellspacing="0" class="itemlist">
            {papers_html}
        </table>
        """

    def next_page(self):
        if self.current_page < self.total_pages:
            self.current_page += 1
        return self.render_papers()

    def prev_page(self):
        if self.current_page > 1:
            self.current_page -= 1
        return self.render_papers()

    def set_time_filter(self, time_filter):
        """
        Set the time filter and fetch papers accordingly.
        """
        if self.fetch_papers(time_filter):
            return self.render_papers()
        else:
            return "<div class='no-papers'>Failed to fetch papers. Please try again later.</div>"

paper_manager = PaperManager()

def initialize_app():
    if paper_manager.fetch_papers():
        return paper_manager.render_papers()
    else:
        return "<div class='no-papers'>Failed to fetch papers. Please try again later.</div>"

def refresh_papers():
    if paper_manager.fetch_papers(paper_manager.time_filter):
        return paper_manager.render_papers()
    else:
        return "<div class='no-papers'>Failed to refresh papers. Please try again later.</div>"

css = """
body {
    background-color: white;
    font-family: Verdana, Geneva, sans-serif;
    margin: 0;
    padding: 0;
}

a {
    color: #0000ff;
    text-decoration: none;
}

a:visited {
    color: #551A8B;
}

.container {
    width: 85%;
    margin: auto;
}

table {
    width: 100%;
}

.header-table {
    width: 100%;
    background-color: #ff6600;
    padding: 2px 10px;
}

.header-table a {
    color: black;
    font-weight: bold;
    font-size: 14pt;
    text-decoration: none;
}

.itemlist .athing {
    background-color: #f6f6ef;
}

.rank {
    font-size: 14pt;
    color: #828282;
    padding-right: 5px;
}

.storylink {
    font-size: 10pt;
}

.subtext {
    font-size: 8pt;
    color: #828282;
    padding-left: 40px;
}

.subtext a {
    color: #828282;
    text-decoration: none;
}

#refresh-button {
    background: none;
    border: none;
    color: black;
    font-weight: bold;
    font-size: 14pt;
    cursor: pointer;
}

.no-papers {
    text-align: center;
    color: #828282;
    padding: 1rem;
    font-size: 14pt;
}

@media (max-width: 640px) {
    .header-table a {
        font-size: 12pt;
    }

    .storylink {
        font-size: 9pt;
    }

    .subtext {
        font-size: 7pt;
    }
}

/* Dark mode */
@media (prefers-color-scheme: dark) {
    body {
        background-color: #121212;
        color: #e0e0e0;
    }

    a {
        color: #add8e6;
    }

    a:visited {
        color: #9370db;
    }

    .header-table {
        background-color: #ff6600;
    }

    .header-table a {
        color: black;
    }

    .itemlist .athing {
        background-color: #1e1e1e;
    }

    .rank {
        color: #b0b0b0;
    }

    .subtext {
        color: #b0b0b0;
    }

    .subtext a {
        color: #b0b0b0;
    }

    #refresh-button {
        color: #e0e0e0;
    }

    .no-papers {
        color: #b0b0b0;
    }
}
"""

demo = gr.Blocks(css=css)

with demo:
    with gr.Column(elem_classes=["container"]):
        # Accordion for Submission Instructions
        with gr.Accordion("How to Submit a Paper", open=False):
            gr.Markdown("""
            ### Steps to Submit Your Paper

            **Step 1:** Search for your paper and index on Hugging Face:  
            [https://huggingface.co/papers?search=true](https://huggingface.co/papers?search=true)

            **Step 2:** Submit the paper to Daily Papers:  
            [https://huggingface.co/papers](https://huggingface.co/papers)

            Once your paper is submitted, it will automatically appear in this demo.
            """)
        
        # Header with Refresh Button and Time Filter
        with gr.Row():
            gr.HTML("""
            <table border="0" cellpadding="0" cellspacing="0" class="header-table">
                <tr>
                    <td>
                        <span class="pagetop">
                            <b class="hnname"><a href="#">Daily Papers</a></b>
                        </span>
                    </td>
                    <td align="right">
                        <button id="refresh-button">Refresh</button>
                    </td>
                </tr>
            </table>
            """)
        
        # Time Filter Dropdown
        with gr.Row(elem_classes=["time-filter-row"], elem_id="time-filter-row"):
            gr.HTML("<label for='time-filter'>Filter by Timeframe: </label>")
            time_filter_dropdown = gr.Dropdown(
                choices=["All Time", "Last Week", "Last Month", "Last Year"],
                value="All Time",
                label="Timeframe",
                interactive=True,
                elem_id="time-filter-dropdown"
            )
        
        # Paper list
        paper_list = gr.HTML()
        
        # Navigation Buttons
        with gr.Row():
            prev_button = gr.Button("Prev")
            next_button = gr.Button("Next")

    # Load papers on app start
    demo.load(initialize_app, outputs=[paper_list])

    # Button clicks
    prev_button.click(paper_manager.prev_page, outputs=[paper_list])
    next_button.click(paper_manager.next_page, outputs=[paper_list])
    refresh_button = gr.Button("Refresh", visible=False, elem_id="refresh-hidden")
    refresh_button.click(refresh_papers, outputs=[paper_list])

    # Time Filter change
    time_filter_dropdown.change(
        paper_manager.set_time_filter,
        inputs=[time_filter_dropdown],
        outputs=[paper_list]
    )

    # Bind the visible Refresh button to the hidden one using JavaScript
    gr.HTML("""
    <script>
    document.getElementById('refresh-button').addEventListener('click', function() {
        document.getElementById('refresh-hidden').click();
    });
    </script>
    """)

demo.launch()