Create app.py

Added a functionality to allow the user to choose between hot papers (computed using hot score) and top papers, which chooses the papers with the most upvotes in a certain time frame (day/week/month year)

Note
1. I wasn't able to run locally https://huggingface.co/api/daily_papers?limit={} with a value large than 100. Please validate it works with a 1000
2. We might need to significantly increase the number of fetched papers to retrieve the top papers from the last year, or we could design some caching mechanism.

app.py

@@ -1,6 +1,6 @@
 import gradio as gr
 import requests
-from datetime import datetime, timezone
+from datetime import datetime, timezone, timedelta
 
 API_URL = "https://huggingface.co/api/daily_papers"
 
@@ -10,8 +10,10 @@ class PaperManager:
         self.current_page = 1
         self.papers = []
         self.total_pages = 1
+        self.sort_method = "hot"
+        self.time_period = "day"
 
-    def calculate_score(self, paper):
+    def calculate_hot_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.
@@ -21,28 +23,53 @@ class PaperManager:
         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
+        time_diff_hours = time_diff.total_seconds() / 3600
 
-        # Avoid division by zero and apply the hotness formula
         score = upvotes / ((time_diff_hours + 2) ** 1.5)
         return score
 
-    def fetch_papers(self):
+    def fetch_papers(self, sort_method, time_period=None):
         try:
-            response = requests.get(f"{API_URL}?limit=100")
+            response = requests.get(f"{API_URL}?limit=1000")  # Increased limit to get more papers
             response.raise_for_status()
             data = response.json()
 
-            # Sort papers by calculated score descending
-            self.papers = sorted(
-                data,
-                key=lambda x: self.calculate_score(x),
-                reverse=True
-            )
+            self.sort_method = sort_method
+            self.time_period = time_period if sort_method == "top" else None
+
+            if sort_method == "hot":
+                self.papers = sorted(
+                    data,
+                    key=lambda x: self.calculate_hot_score(x),
+                    reverse=True
+                )
+            elif sort_method == "top":
+                current_time = datetime.now(timezone.utc)
+
+                if time_period == "day":
+                    start_time = current_time - timedelta(days=1)
+                elif time_period == "week":
+                    start_time = current_time - timedelta(days=7)
+                elif time_period == "month":
+                    start_time = current_time - timedelta(days=30)
+                elif time_period == "year":
+                    start_time = current_time - timedelta(days=365)
+                else:
+                    start_time = datetime.min.replace(tzinfo=timezone.utc)
+
+                filtered_papers = [
+                    paper for paper in data
+                    if datetime.fromisoformat(paper.get('publishedAt', '').replace('Z', '+00:00')) >= start_time
+                ]
+
+                self.papers = sorted(
+                    filtered_papers,
+                    key=lambda x: x.get('paper', {}).get('upvotes', 0),
+                    reverse=True
+                )
 
             self.total_pages = max((len(self.papers) + self.papers_per_page - 1) // self.papers_per_page, 1)
             self.current_page = 1
@@ -112,17 +139,12 @@ class PaperManager:
 
 paper_manager = PaperManager()
 
-def initialize_app():
-    if paper_manager.fetch_papers():
-        return paper_manager.render_papers()
+def update_papers(sort_method, time_period=None):
+    if paper_manager.fetch_papers(sort_method, time_period):
+        return paper_manager.render_papers(), gr.update(visible=(sort_method == "top"))
     else:
-        return "<div class='no-papers'>Failed to fetch papers. Please try again later.</div>"
+        return "<div class='no-papers'>Failed to fetch papers. Please try again later.</div>", gr.update(visible=False)
 
-def refresh_papers():
-    if paper_manager.fetch_papers():
-        return paper_manager.render_papers()
-    else:
-        return "<div class='no-papers'>Failed to refresh papers. Please try again later.</div>"
 
 css = """
 body {
@@ -131,63 +153,51 @@ body {
     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;
@@ -196,71 +206,56 @@ table {
     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;
     }
@@ -285,6 +280,7 @@ with demo:
             Once your paper is submitted, it will automatically appear in this demo.
             """)
         # Header with Refresh Button
+
         with gr.Row():
             gr.HTML("""
             <table border="0" cellpadding="0" cellspacing="0" class="header-table">
@@ -300,22 +296,44 @@ with demo:
                 </tr>
             </table>
             """)
+        
+        # Sorting method selection
+        with gr.Row():
+            sort_method = gr.Radio(
+                choices=["hot", "top"],
+                value="hot",
+                label="Sort by:",
+                interactive=True
+            )
+
+            # Time Period Dropdown (only visible when "top" is selected)
+            time_period = gr.Dropdown(
+                choices=["day", "week", "month", "year"],
+                value="day",
+                label="Top papers from the last:",
+                visible=False
+            )
+
         # 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])
+    demo.load(lambda: update_papers("hot"), outputs=[paper_list, time_period])
+
+    # Update papers when sorting method or time period changes
+    sort_method.change(update_papers, inputs=[sort_method, time_period], outputs=[paper_list, time_period])
+    time_period.change(lambda s, t: update_papers(s, t), inputs=[sort_method, time_period], outputs=[paper_list, time_period])
 
     # 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])
+    refresh_button.click(update_papers, inputs=[sort_method, time_period], outputs=[paper_list, time_period])
 
     # Bind the visible Refresh button to the hidden one using JavaScript
     gr.HTML("""