Update app.py

app.py

@@ -5,8 +5,9 @@ from datetime import datetime, timedelta
 import time
 import os
 
-def get_contribution_count(username, github_token, days=30):
-    """Get accurate contribution count using GitHub GraphQL API"""
+
+def get_contribution_count(username, github_token, days=365):
+    """Fetch contribution count using GitHub GraphQL API"""
     if not github_token:
         return 0
     
@@ -49,73 +50,60 @@ def get_contribution_count(username, github_token, days=30):
             if 'data' in data and data['data'] and data['data']['user']:
                 return data['data']['user']['contributionsCollection']['contributionCalendar']['totalContributions']
     except Exception as e:
-        print(f"Error fetching contributions for {username}: {e}")
+        print(f"Error for {username}: {e}")
     
     return 0
 
-def fetch_github_users(github_token=None, max_users=200, min_followers=0, days=30):
-    """
-    Fetch GitHub users from UAE using committers.top methodology:
-    1. Search by location + sort by followers
-    2. Get top N users by followers
-    3. Get their contributions
-    4. Re-sort by contributions
-    """
+
+def fetch_github_users(github_token=None, max_users=256, min_followers=34, days=365):
+    """Fetch UAE GitHub users matching committers.top methodology"""
     if not github_token:
         github_token = os.getenv('GITHUB_TOKEN')
     
     if not github_token:
-        return pd.DataFrame(), "❌ GitHub token is REQUIRED for GraphQL API (contribution counting)"
-
+        return pd.DataFrame(), "GitHub token is required"
+    
     headers = {
         'Accept': 'application/vnd.github.v3+json',
         'Authorization': f'token {github_token}'
     }
-
-    all_users = {}  # Use dict to avoid duplicates
+    
+    all_users = {}
     status_updates = []
     api_calls = 0
-
-    # UAE location search terms
-    locations = [
-        'United Arab Emirates', 'UAE',
-        'Dubai', 'Abu Dhabi', 'Sharjah', 'Ajman',
-        'Ras Al Khaimah', 'Fujairah', 'Umm Al Quwain'
-    ]
-
+    
+    uae_keywords = ['dubai', 'abu dhabi', 'abudhabi', 'sharjah', 'ajman',
+                    'uae', 'u.a.e', 'united arab emirates', 'ras al khaimah',
+                    'fujairah', 'umm al quwain', 'emirates', 'rak']
+    
     try:
-        status_updates.append(f"🔍 Step 1: Searching for UAE users (sorted by followers)...")
+        status_updates.append(f"Searching UAE users (min {min_followers} followers)...")
         
-        # STEP 1: Get users sorted by FOLLOWERS (like committers.top does)
-        for location in locations:
-            if api_calls >= 100:
-                break
-            
-            # Search by followers DESC
-            search_url = f'https://api.github.com/search/users?q=location:"{location}"+followers:>={min_followers}&sort=followers&order=desc&per_page=100'
-            
-            response = requests.get(search_url, headers=headers)
-            api_calls += 1
+        # Search by location - get top 1000 by followers
+        search_url = f'https://api.github.com/search/users?q=location:"United Arab Emirates"+followers:>={min_followers}&sort=followers&order=desc&per_page=100'
+        
+        response = requests.get(search_url, headers=headers)
+        api_calls += 1
+        
+        if response.status_code == 200:
+            data = response.json()
+            total_count = data.get('total_count', 0)
+            users = data.get('items', [])
+            status_updates.append(f"Found {total_count} UAE users, fetching details...")
             
-            if response.status_code == 200:
-                data = response.json()
-                users = data.get('items', [])
-                status_updates.append(f"  📍 {location}: {len(users)} users")
-                
+            page = 1
+            while users and len(all_users) < 1000 and api_calls < 150:
                 for user in users:
-                    if user['login'] not in all_users and api_calls < 100:
-                        # Get detailed user info
+                    if len(all_users) >= 1000:
+                        break
+                    
+                    if user['login'] not in all_users:
                         user_response = requests.get(user['url'], headers=headers)
                         api_calls += 1
                         
                         if user_response.status_code == 200:
                             user_data = user_response.json()
-                            
-                            # Verify UAE location (case insensitive)
                             user_location = (user_data.get('location') or '').lower()
-                            uae_keywords = ['dubai', 'abu dhabi', 'abudhabi', 'sharjah', 'ajman', 
-                                           'uae', 'u.a.e', 'united arab emirates', 'ras al khaimah', 
-                                           'fujairah', 'umm al quwain', 'emirates', 'rak']
                             
                             if any(kw in user_location for kw in uae_keywords):
                                 all_users[user['login']] = {
@@ -125,48 +113,61 @@ def fetch_github_users(github_token=None, max_users=200, min_followers=0, days=3
                                     'followers': user_data.get('followers', 0),
                                     'public_repos': user_data.get('public_repos', 0),
                                     'location': user_data.get('location', ''),
-                                    'contributions': 0  # Will fill later
+                                    'contributions': 0
                                 }
                         
                         time.sleep(0.5)
                 
+                if len(users) < 100:
+                    break
+                
+                page += 1
+                if page > 10:
+                    break
+                
+                next_url = f'https://api.github.com/search/users?q=location:"United Arab Emirates"+followers:>={min_followers}&sort=followers&order=desc&per_page=100&page={page}'
+                response = requests.get(next_url, headers=headers)
+                api_calls += 1
+                
+                if response.status_code == 200:
+                    users = response.json().get('items', [])
+                    status_updates.append(f"  Page {page}: fetched")
+                else:
+                    break
+                
                 time.sleep(1)
-            
-            elif response.status_code == 403:
-                status_updates.append(f"⚠️ Rate limit hit")
-                break
         
-        status_updates.append(f"✅ Found {len(all_users)} unique UAE users")
+        elif response.status_code == 403:
+            return pd.DataFrame(), "Rate limit exceeded"
+        
+        status_updates.append(f"Collected {len(all_users)} UAE users (top 1000 by followers)")
         
-        # STEP 2: Get contributions for collected users
-        status_updates.append(f"📊 Step 2: Fetching contribution counts...")
+        # Get contributions for all collected users
+        status_updates.append(f"Fetching contributions (last {days} days)...")
         
         count = 0
         for login, user in all_users.items():
-            if api_calls >= 200:
+            if api_calls >= 300:
                 break
             
             contributions = get_contribution_count(login, github_token, days)
             user['contributions'] = contributions
             
             count += 1
-            if count % 20 == 0:
-                status_updates.append(f"  ⏳ Processed {count}/{len(all_users)} users...")
+            if count % 50 == 0:
+                status_updates.append(f"  Progress: {count}/{len(all_users)} users...")
             
             time.sleep(0.5)
         
-        # STEP 3: Sort by CONTRIBUTIONS (like committers.top final step)
+        # Sort by contributions
         users_list = list(all_users.values())
         users_list.sort(key=lambda x: (x['contributions'], x['followers']), reverse=True)
         
-        # Take top N
         top_users = users_list[:max_users]
         
-        # Add rank
         for i, user in enumerate(top_users, 1):
             user['rank'] = i
         
-        # Create DataFrame
         df = pd.DataFrame(top_users)
         
         if not df.empty:
@@ -174,27 +175,20 @@ def fetch_github_users(github_token=None, max_users=200, min_followers=0, days=3
             display_df.columns = ['Rank', 'Name', 'Username', 'Contributions', 'Followers', 'Public Repos', 'Location']
             display_df['GitHub Profile'] = df['login'].apply(lambda x: f"https://github.com/{x}")
             
-            status_message = f"""✅ Success! Top {len(df)} UAE contributors (last {days} days)
-📊 Total UAE users found: {len(all_users)}
-🔧 API calls used: {api_calls}
-
-Methodology (like committers.top):
-1. Search by location + sort by followers
-2. Get top users by followers  
-3. Fetch contribution counts
-4. Re-sort by contributions
-
-""" + "\n".join(status_updates[-5:])
+            status_message = f"Success! Top {len(df)} UAE contributors (last {days} days)\n"
+            status_message += f"Total UAE users collected: {len(all_users)}, API calls: {api_calls}\n\n"
+            status_message += "\n".join(status_updates[-5:])
             
             return display_df, status_message
         else:
-            return pd.DataFrame(), "⚠️ No users found"
+            return pd.DataFrame(), "No users found"
     
     except Exception as e:
-        return pd.DataFrame(), f"❌ Error: {str(e)}"
+        return pd.DataFrame(), f"Error: {str(e)}"
+
 
 def search_users(df, search_term):
-    """Filter users"""
+    """Filter users by name or username"""
     if df is None or df.empty or not search_term:
         return df
     
@@ -205,30 +199,23 @@ def search_users(df, search_term):
     )
     return df[mask]
 
-# Gradio interface
-with gr.Blocks(theme=gr.themes.Soft(), title="GitHub UAE Top Contributors") as app:
 
-    gr.Markdown("""
-    # 🇦🇪 GitHub UAE Top Contributors
-    ### Using committers.top Methodology
-    
-    **Approach:** Search by location → Sort by followers → Get contributions → Re-sort by contributions
+with gr.Blocks(theme=gr.themes.Soft(), title="GitHub UAE Top Contributors") as app:
+    gr.Markdown("# GitHub UAE Top Contributors")
+    gr.Markdown("*Following committers.top methodology*")
     
-    **Reference:** [committers.top/uae](https://committers.top/uae.html) found 31,463 UAE users!
-    """)
-
     with gr.Row():
         token_input = gr.Textbox(
-            label="GitHub Token (REQUIRED - needs 'read:user' scope)",
-            placeholder="ghp_xxxx or set GITHUB_TOKEN env var",
+            label="GitHub Token",
+            placeholder="ghp_xxxx",
             type="password",
             scale=3
         )
         max_users_input = gr.Slider(
             label="Max Users to Display",
-            minimum=10,
+            minimum=50,
             maximum=500,
-            value=200,
+            value=256,
             step=10,
             scale=1
         )
@@ -236,26 +223,26 @@ with gr.Blocks(theme=gr.themes.Soft(), title="GitHub UAE Top Contributors") as a
     with gr.Row():
         min_followers_input = gr.Slider(
             label="Min Followers (committers.top uses 34)",
-            minimum=0,
+            minimum=10,
             maximum=100,
-            value=0,
+            value=34,
             step=1
         )
         days_input = gr.Slider(
-            label="Days to Count",
-            minimum=7,
-            maximum=90,
-            value=30,
+            label="Days to Count (committers.top uses 365)",
+            minimum=30,
+            maximum=365,
+            value=365,
             step=1
         )
 
-    fetch_btn = gr.Button("🚀 Fetch Top Contributors", variant="primary", size="lg")
+    fetch_btn = gr.Button("Fetch Top Contributors", variant="primary", size="lg")
     
     status_msg = gr.Textbox(label="Status", interactive=False, lines=6)
 
     with gr.Row():
         search_box = gr.Textbox(
-            label="🔍 Filter Results",
+            label="Filter Results",
             placeholder="Search by name or username...",
             scale=4
         )
@@ -270,39 +257,6 @@ with gr.Blocks(theme=gr.themes.Soft(), title="GitHub UAE Top Contributors") as a
         interactive=False
     )
 
-    gr.Markdown("""
-    ---
-    ## How This Works:
-    
-    **Methodology (same as committers.top):**
-    ```
-    1. Search: location:UAE + sort:followers
-    2. Filter: Verify location has UAE keywords
-    3. Collect: Top N users by followers
-    4. Count: Get contributions via GraphQL
-    5. Sort: Re-rank by contributions
-    ```
-    
-    **Why This Finds Everyone:**
-    - Searches multiple UAE locations
-    - Sorts by followers first (GitHub's default)
-    - Gets top users who are likely active
-    - Then accurately counts contributions
-    - Re-sorts by actual contributions
-    
-    **Token Setup:**
-    1. Go to: https://github.com/settings/tokens
-    2. Generate new token (classic)
-    3. **Must have `read:user` scope** (for GraphQL)
-    4. Copy and paste above
-    
-    **Expected Results:**
-    - Should find: tschm (~10k), zning1994 (~9k), YASSERRMD (~788), nkapila6 (~446)
-    - All users from committers.top list should appear
-    
-    **Note:** Takes 2-5 minutes. Be patient!
-    """)
-
     def fetch_and_display(token, max_users, min_followers, days):
         df, msg = fetch_github_users(token or None, int(max_users), int(min_followers), int(days))
         return df, df, msg
@@ -331,5 +285,6 @@ with gr.Blocks(theme=gr.themes.Soft(), title="GitHub UAE Top Contributors") as a
         outputs=[search_box, data_display]
     )
 
+
 if __name__ == "__main__":
-    app.launch()
+    app.launch()