Spaces:
Sleeping
Sleeping
Create app.py
Browse files
app.py
ADDED
|
@@ -0,0 +1,245 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import gradio as gr
|
| 2 |
+
import requests
|
| 3 |
+
import pandas as pd
|
| 4 |
+
from datetime import datetime, timedelta
|
| 5 |
+
import time
|
| 6 |
+
import os
|
| 7 |
+
|
| 8 |
+
def fetch_github_users(github_token=None, max_users=256, min_followers=34):
|
| 9 |
+
"""
|
| 10 |
+
Fetch GitHub users from UAE directly using GitHub API
|
| 11 |
+
"""
|
| 12 |
+
# Read token from environment variable if not provided
|
| 13 |
+
if not github_token:
|
| 14 |
+
github_token = os.getenv('GITHUB_TOKEN')
|
| 15 |
+
|
| 16 |
+
headers = {
|
| 17 |
+
'Accept': 'application/vnd.github.v3+json',
|
| 18 |
+
}
|
| 19 |
+
|
| 20 |
+
if github_token:
|
| 21 |
+
headers['Authorization'] = f'token {github_token}'
|
| 22 |
+
|
| 23 |
+
all_users = []
|
| 24 |
+
|
| 25 |
+
# Search locations in UAE
|
| 26 |
+
locations = [
|
| 27 |
+
'Dubai', 'Abu Dhabi', 'Sharjah', 'Ajman',
|
| 28 |
+
'United Arab Emirates', 'UAE',
|
| 29 |
+
'Ras Al Khaimah', 'Fujairah', 'Umm Al Quwain'
|
| 30 |
+
]
|
| 31 |
+
|
| 32 |
+
status_updates = []
|
| 33 |
+
|
| 34 |
+
try:
|
| 35 |
+
for location in locations:
|
| 36 |
+
status_updates.append(f"π Searching users in {location}...")
|
| 37 |
+
|
| 38 |
+
# Search users by location and followers
|
| 39 |
+
search_url = f'https://api.github.com/search/users?q=location:{location}+followers:>={min_followers}&sort=followers&order=desc&per_page=100'
|
| 40 |
+
|
| 41 |
+
response = requests.get(search_url, headers=headers)
|
| 42 |
+
|
| 43 |
+
if response.status_code == 200:
|
| 44 |
+
data = response.json()
|
| 45 |
+
users = data.get('items', [])
|
| 46 |
+
|
| 47 |
+
for user in users:
|
| 48 |
+
if not any(u['login'] == user['login'] for u in all_users):
|
| 49 |
+
# Fetch detailed user info
|
| 50 |
+
user_url = user['url']
|
| 51 |
+
user_response = requests.get(user_url, headers=headers)
|
| 52 |
+
|
| 53 |
+
if user_response.status_code == 200:
|
| 54 |
+
user_data = user_response.json()
|
| 55 |
+
|
| 56 |
+
# Fetch user's contribution data (public repos and events)
|
| 57 |
+
events_url = f"https://api.github.com/users/{user['login']}/events/public"
|
| 58 |
+
events_response = requests.get(events_url, headers=headers)
|
| 59 |
+
|
| 60 |
+
contributions = 0
|
| 61 |
+
if events_response.status_code == 200:
|
| 62 |
+
events = events_response.json()
|
| 63 |
+
# Count push events as contributions
|
| 64 |
+
contributions = len([e for e in events if e.get('type') == 'PushEvent'])
|
| 65 |
+
|
| 66 |
+
all_users.append({
|
| 67 |
+
'login': user_data.get('login', ''),
|
| 68 |
+
'name': user_data.get('name', user_data.get('login', '')),
|
| 69 |
+
'avatar': user_data.get('avatar_url', ''),
|
| 70 |
+
'followers': user_data.get('followers', 0),
|
| 71 |
+
'public_repos': user_data.get('public_repos', 0),
|
| 72 |
+
'contributions': contributions + user_data.get('public_repos', 0),
|
| 73 |
+
'location': user_data.get('location', ''),
|
| 74 |
+
'bio': user_data.get('bio', ''),
|
| 75 |
+
'company': user_data.get('company', ''),
|
| 76 |
+
})
|
| 77 |
+
|
| 78 |
+
time.sleep(0.5) # Rate limiting
|
| 79 |
+
|
| 80 |
+
if len(all_users) >= max_users:
|
| 81 |
+
break
|
| 82 |
+
|
| 83 |
+
status_updates.append(f"β
Found {len(users)} users in {location}")
|
| 84 |
+
time.sleep(1) # Rate limiting between searches
|
| 85 |
+
|
| 86 |
+
elif response.status_code == 403:
|
| 87 |
+
status_updates.append(f"β οΈ Rate limit reached. Please add a GitHub token.")
|
| 88 |
+
break
|
| 89 |
+
else:
|
| 90 |
+
status_updates.append(f"β Error searching {location}: {response.status_code}")
|
| 91 |
+
|
| 92 |
+
if len(all_users) >= max_users:
|
| 93 |
+
break
|
| 94 |
+
|
| 95 |
+
# Sort by followers and contributions
|
| 96 |
+
all_users.sort(key=lambda x: (x['followers'], x['contributions']), reverse=True)
|
| 97 |
+
|
| 98 |
+
# Add rank
|
| 99 |
+
for i, user in enumerate(all_users[:max_users], 1):
|
| 100 |
+
user['rank'] = i
|
| 101 |
+
|
| 102 |
+
# Convert to DataFrame
|
| 103 |
+
df = pd.DataFrame(all_users[:max_users])
|
| 104 |
+
|
| 105 |
+
if not df.empty:
|
| 106 |
+
display_df = df[['rank', 'name', 'login', 'followers', 'public_repos', 'contributions', 'location']].copy()
|
| 107 |
+
display_df.columns = ['Rank', 'Name', 'Username', 'Followers', 'Public Repos', 'Contributions', 'Location']
|
| 108 |
+
display_df['GitHub Profile'] = df['login'].apply(lambda x: f"https://github.com/{x}")
|
| 109 |
+
|
| 110 |
+
status_message = f"β
Successfully fetched {len(df)} users\n" + "\n".join(status_updates[-5:])
|
| 111 |
+
return display_df, status_message
|
| 112 |
+
else:
|
| 113 |
+
return pd.DataFrame(), "β οΈ No users found\n" + "\n".join(status_updates)
|
| 114 |
+
|
| 115 |
+
except Exception as e:
|
| 116 |
+
return pd.DataFrame(), f"β Error: {str(e)}\n" + "\n".join(status_updates)
|
| 117 |
+
|
| 118 |
+
def search_users(df, search_term):
|
| 119 |
+
"""Filter users based on search term"""
|
| 120 |
+
if df is None or df.empty:
|
| 121 |
+
return df
|
| 122 |
+
|
| 123 |
+
if not search_term:
|
| 124 |
+
return df
|
| 125 |
+
|
| 126 |
+
search_term = search_term.lower()
|
| 127 |
+
mask = (
|
| 128 |
+
df['Name'].str.lower().str.contains(search_term, na=False) |
|
| 129 |
+
df['Username'].str.lower().str.contains(search_term, na=False)
|
| 130 |
+
)
|
| 131 |
+
return df[mask]
|
| 132 |
+
|
| 133 |
+
# Create Gradio interface
|
| 134 |
+
with gr.Blocks(theme=gr.themes.Soft(), title="GitHub UAE Users Fetcher") as app:
|
| 135 |
+
|
| 136 |
+
gr.Markdown("""
|
| 137 |
+
# π GitHub Users in UAE - Direct Fetcher
|
| 138 |
+
### Fetch active GitHub users directly from GitHub API
|
| 139 |
+
|
| 140 |
+
**Note:** The app will automatically use `GITHUB_TOKEN` environment variable if set.
|
| 141 |
+
Without a token, you're limited to 60 requests/hour. With a token: 5000 requests/hour.
|
| 142 |
+
""")
|
| 143 |
+
|
| 144 |
+
with gr.Row():
|
| 145 |
+
token_input = gr.Textbox(
|
| 146 |
+
label="GitHub Personal Access Token (Optional - overrides env variable)",
|
| 147 |
+
placeholder="ghp_xxxxxxxxxxxx or leave empty to use GITHUB_TOKEN env var",
|
| 148 |
+
type="password",
|
| 149 |
+
scale=3
|
| 150 |
+
)
|
| 151 |
+
max_users_input = gr.Slider(
|
| 152 |
+
label="Max Users",
|
| 153 |
+
minimum=10,
|
| 154 |
+
maximum=500,
|
| 155 |
+
value=256,
|
| 156 |
+
step=1,
|
| 157 |
+
scale=1
|
| 158 |
+
)
|
| 159 |
+
|
| 160 |
+
with gr.Row():
|
| 161 |
+
min_followers_input = gr.Slider(
|
| 162 |
+
label="Minimum Followers",
|
| 163 |
+
minimum=1,
|
| 164 |
+
maximum=100,
|
| 165 |
+
value=34,
|
| 166 |
+
step=1
|
| 167 |
+
)
|
| 168 |
+
|
| 169 |
+
with gr.Row():
|
| 170 |
+
fetch_btn = gr.Button("π Fetch Users from GitHub", variant="primary", size="lg")
|
| 171 |
+
|
| 172 |
+
status_msg = gr.Textbox(label="Status", interactive=False, lines=3)
|
| 173 |
+
|
| 174 |
+
with gr.Row():
|
| 175 |
+
search_box = gr.Textbox(
|
| 176 |
+
label="π Search by Name or Username",
|
| 177 |
+
placeholder="Type to search...",
|
| 178 |
+
scale=4
|
| 179 |
+
)
|
| 180 |
+
clear_btn = gr.Button("Clear", scale=1)
|
| 181 |
+
|
| 182 |
+
# Store the full dataframe
|
| 183 |
+
full_data = gr.State(value=pd.DataFrame())
|
| 184 |
+
|
| 185 |
+
# Display dataframe
|
| 186 |
+
data_display = gr.Dataframe(
|
| 187 |
+
headers=["Rank", "Name", "Username", "Followers", "Public Repos", "Contributions", "Location", "GitHub Profile"],
|
| 188 |
+
datatype=["number", "str", "str", "number", "number", "number", "str", "str"],
|
| 189 |
+
wrap=True,
|
| 190 |
+
interactive=False
|
| 191 |
+
)
|
| 192 |
+
|
| 193 |
+
gr.Markdown("""
|
| 194 |
+
---
|
| 195 |
+
**How to set up GitHub Token:**
|
| 196 |
+
|
| 197 |
+
**Option 1: Environment Variable (Recommended)**
|
| 198 |
+
```bash
|
| 199 |
+
export GITHUB_TOKEN="ghp_your_token_here"
|
| 200 |
+
python app.py
|
| 201 |
+
```
|
| 202 |
+
|
| 203 |
+
**Option 2: Manual Input**
|
| 204 |
+
1. Go to [GitHub Settings > Developer settings > Personal access tokens](https://github.com/settings/tokens)
|
| 205 |
+
2. Generate new token (classic)
|
| 206 |
+
3. No special scopes needed for public data
|
| 207 |
+
4. Copy and paste the token in the field above
|
| 208 |
+
|
| 209 |
+
**Data fetched directly from:** GitHub API
|
| 210 |
+
""")
|
| 211 |
+
|
| 212 |
+
# Event handlers
|
| 213 |
+
def fetch_and_display(token, max_users, min_followers):
|
| 214 |
+
df, msg = fetch_github_users(token if token else None, int(max_users), int(min_followers))
|
| 215 |
+
return df, df, msg
|
| 216 |
+
|
| 217 |
+
def filter_data(df, search):
|
| 218 |
+
if df is None or df.empty:
|
| 219 |
+
return df
|
| 220 |
+
return search_users(df, search)
|
| 221 |
+
|
| 222 |
+
def clear_search(df):
|
| 223 |
+
return "", df
|
| 224 |
+
|
| 225 |
+
fetch_btn.click(
|
| 226 |
+
fn=fetch_and_display,
|
| 227 |
+
inputs=[token_input, max_users_input, min_followers_input],
|
| 228 |
+
outputs=[full_data, data_display, status_msg]
|
| 229 |
+
)
|
| 230 |
+
|
| 231 |
+
search_box.change(
|
| 232 |
+
fn=filter_data,
|
| 233 |
+
inputs=[full_data, search_box],
|
| 234 |
+
outputs=data_display
|
| 235 |
+
)
|
| 236 |
+
|
| 237 |
+
clear_btn.click(
|
| 238 |
+
fn=clear_search,
|
| 239 |
+
inputs=[full_data],
|
| 240 |
+
outputs=[search_box, data_display]
|
| 241 |
+
)
|
| 242 |
+
|
| 243 |
+
# Launch the app
|
| 244 |
+
if __name__ == "__main__":
|
| 245 |
+
app.launch()
|