Skip to content

python_sandbox.md

Latest commit

 

History

History
120 lines (86 loc) · 3.36 KB

python_sandbox.md

File metadata and controls

120 lines (86 loc) · 3.36 KB
Raw

Python Sandbox Guide

Python Sandbox lets you create graph images and math animation outputs from Python code, then save or insert the generated files into the editor.

Open it from Menu Sidebar > Python Sandbox.

What It Is Best For

  • Matplotlib graphs
  • Data visualizations
  • Transparent PNG chart assets
  • Manim math animations
  • Educational visual elements for explainer videos

How To Use

  1. Open a project.
  2. Open Python Sandbox from the menu sidebar.
  3. Choose Python Script or Manim Scene.
  4. Write code in the editable code panel.
  5. Click Run.
  6. Preview the generated output.
  7. Use Save to Assets or Insert.

Generated files are detected automatically when they are saved inside the provided out folder.

Sandbox Environment

Released builds include a bundled Python Sandbox runtime with the required packages already installed: NumPy, Matplotlib, Pillow, SciPy, ImageIO, ImageIO FFmpeg support, and Manim. Users should not need to install these packages on their computer.

For local development or recovery, the app can still create a local repair sandbox in the app data folder when a bundled runtime is not available. Release builds should prepare the bundled runtime before packaging:

npm run prepare-python-sandbox

Then package the app normally:

npm run package

Preloaded Libraries

The sandbox preloads common libraries and variables. You should use these directly instead of writing import lines.

Available names include:

np
plt
animation
math
random
statistics
Path
out
WIDTH
HEIGHT
FPS

For Manim scenes, common Manim names are also preloaded when Manim is installed.

Important Rule

Do not write import lines in the editable code area.

Blocked examples:

import numpy as np
from manim import *

The sandbox already provides the supported libraries. If an import line is present, Run will be disabled or rejected until you remove it.

Matplotlib Example

x = np.linspace(-5, 5, 300)
y = x ** 2

plt.figure(figsize=(10, 5.625), dpi=160)
plt.plot(x, y, color="#8b5cf6", linewidth=4)
plt.grid(True, alpha=0.25)
plt.tight_layout()
plt.savefig(os.path.join(out, "parabola.png"), transparent=True)

Manim Example

Use the scene class name shown in the Scene class field.

class GeneratedScene(Scene):
    def construct(self):
        title = Text("Huber Loss", font_size=48)
        formula = MathTex(r"L_\delta(a)")
        formula.next_to(title, DOWN, buff=0.5)

        self.play(Write(title))
        self.play(Write(formula))
        self.wait(1)

Getting Better Results

  • Save outputs inside out.
  • Use clear filenames such as loss_curve.png or gradient_descent.mp4.
  • Use transparent backgrounds for overlays when possible.
  • Match the project aspect ratio with WIDTH, HEIGHT, and FPS.
  • For graphs, use large figure size and high DPI for sharp exports.
  • For Manim, keep scene class names simple and match the selected class field.
  • Use Insert when you want the generated file placed directly on the current scene.
  • Use Save to Assets when you want to keep the generated file for later.

Safety And Limits

The sandbox blocks import statements, package installation attempts, subprocess calls, and dynamic execution helpers such as eval and exec. This keeps the feature focused on creating editor assets with the supported preloaded tools.