Profiler Powered by Scalene

A powerful VS Code extension for Scalene - the AI-powered CPU, GPU, and memory profiler for Python.

Built for Open VSX - Works in Cursor, VSCodium, and other VS Code alternatives.

Author: Carlo Bertini (WaYdotNET)

🎯 Inspired By

This extension is inspired by the official Scalene VS Code Extension by Emery Berger.

The original extension is available on the VS Code Marketplace but not on Open VSX, which is required for Cursor, VSCodium, and other open-source VS Code alternatives.

✨ What's Different from the Original

Features

Requirements

• Python 3.8+
• Scalene (auto-installed if missing)
• Optional: uv (for faster installation if pyproject.toml exists)

Quick Start

1. Open a Python file
2. Click the 🔥 Profiler button in the status bar, or
3. Right-click → Profiler: Profile Current File
4. View results in the integrated viewer with source code + metrics

Commands

Profile Viewer

The built-in profile viewer uses the Rosé Pine Dawn theme (WCAG AAA compliant) and has four tabs:

📊 Line Profile

Shows for each line:

• Line Number
• 📝 Source Code (the actual code!)
• CPU (Python) - Time in Python code
• CPU (Native) - Time in C/C++ libraries
• Memory (Avg MB) - Average memory allocated
• Memory (Peak MB) - Peak memory usage
• Copy (MB/s) - Memory copy rate (high = inefficient)
• Growth (MB) - Memory growth (leak indicator)

⚡ Functions

• Function name and location
• CPU time breakdown
• Memory usage
• Peak memory

🚨 Potential Leaks

• 📝 Source code of leaking lines
• Leak score (0-100%)
• Memory allocated
• File:Line location

📖 About

• Complete metrics guide and quick tips
• Color coding reference
• Links to Scalene, extension repository, and theme

Extension Settings

Development

npm install       # Install dependencies
npm run compile   # Build
npm run watch     # Watch mode
npm run package   # Create .vsix

Requirements: Node.js 18+, npm

Installation

From Open VSX (Cursor/VSCodium)

Search for "Profiler Powered by Scalene" in the Extensions view, or:

ext install WaYdotNET.profiler-powered-by-scalene

From VSIX

1. Download the .vsix file from Releases
2. In VS Code/Cursor: Ctrl+Shift+P → "Install from VSIX"
3. Select the .vsix file

🐍 Profiling Django Applications

Method 1: Test Client Script (⭐ Recommended)

# profile_api.py
import os, django
os.environ.setdefault('DJANGO_SETTINGS_MODULE', 'myproject.settings')
django.setup()

from django.test import Client
client = Client()

for i in range(100):
    response = client.get('/api/users/')
    print(f"✓ {i}: {response.status_code}")

Run: Profiler: Profile Current File on profile_api.py

Method 2: Profile runserver

scalene run -o profile.json manage.py --- runserver 0:8000 --noreload
# Make requests, then Ctrl+C
# In Cursor: Profiler: View Last Profile

⚠️ Always use --noreload to prevent auto-reloader issues

Method 3: Profile Specific View

# profile_view.py
import os, django
os.environ.setdefault('DJANGO_SETTINGS_MODULE', 'myproject.settings')
django.setup()

from myapp.views import expensive_view
from django.test import RequestFactory

for i in range(50):
    request = RequestFactory().get('/my-url/')
    response = expensive_view(request)

Method 4: Profile Management Command

# profile_command.py
import os
os.environ.setdefault('DJANGO_SETTINGS_MODULE', 'myproject.settings')
from django.core.management import execute_from_command_line

execute_from_command_line(['manage.py', 'your_command', '--args'])

Method 5: Profile ORM Queries

# profile_queries.py
import os, django
os.environ.setdefault('DJANGO_SETTINGS_MODULE', 'myproject.settings')
django.setup()

from myapp.models import User

users = User.objects.all()[:100]
for user in users:
    orders = user.orders.all()  # N+1 query?
    print(f"User {user.id}: {len(orders)} orders")

Look for: High CPU (Native) = DB calls, High Memory (Peak) = Large querysets

🔍 What to Look For

💡 Best Practices

• ✅ Use --noreload with runserver profiling
• ✅ Profile 50-100 iterations for reliable data
• ✅ Use Profiler: Profile Memory Only for leak detection
• ✅ Profile realistic workloads (not empty DBs)

Credits

• Scalene - The underlying profiler by Emery Berger, Sam Stern, and Juan Altmayer Pizzorno
• Original VS Code Extension - Inspiration for this extension by Emery Berger
• Carlo Bertini (WaYdotNET) - Author of this enhanced Open VSX version

License

Apache License 2.0 - see LICENSE

Made with ❤️ by WaYdotNET