Testing & Validation

To ensure that Udility Diffuser is correctly configured and generating high-quality educational illustrations, follow the validation steps outlined below.

Prerequisites for Testing

Before running any tests, verify that your environment is properly configured:

API Connectivity: Ensure you have a valid OpenRouter API key.
Environment Variable: The key must be exported as OPENROUTER_API_KEY.
Dependencies: Verify that cairosvg and matplotlib are installed, as these are required for rendering and displaying the output.

Running the Notebook Tests

The quickest way to validate the installation is via the provided Google Colab Notebook.

Open the notebook in your browser.
Insert your OPENROUTER_API_KEY in the setup cell.
Run the "Usage Examples" cells to verify the end-to-end pipeline (Text → SVG → PNG).

Manual Functional Validation

You can perform a quick smoke test by running a Python script to generate a basic geometric illustration. This validates that the Llama-3.5 model is responding and the SVG conversion engine is operational.

from Udility import diffuser

# Test: Generate a simple scientific diagram
try:
    diffuser.generate_image_from_text(
        "A simple diagram showing a right-angled triangle with labels a, b, and c.",
        output_filename="test_validation.png"
    )
    print("Validation Successful: Image generated and saved.")
except Exception as e:
    print(f"Validation Failed: {e}")

Validating Output Quality

Since Udility Diffuser is designed for educational and illustrative purposes, validation should focus on the following criteria:

File Integrity: Confirm that output.png is created in your working directory and is not a 0-byte file.
Label Legibility: Open the generated PNG to ensure that text labels (generated via SVG <text> tags) are legible and correctly positioned.
SVG-to-PNG Accuracy: Ensure the image contains distinct shapes and colors as requested in the prompt. If the image is blank or corrupted, check if cairosvg dependencies (like libcairo) are missing from your system.
Educational Context: Verify that the "Text-Based Contexting" has correctly interpreted scientific or mathematical concepts (e.g., verifying that a "distance vs. displacement" graph correctly illustrates the two different paths).

Common Validation Errors