Modification / Extension
========================

Front‑End Modifications / Extensions
------------------------------------

1. **Implement Main‑Camera capture in real time**

   - Obtain an **Apple Developer** account.
   - Request for a license to `access the Main Camera API <https://developer.apple.com/documentation/visionos/accessing-the-main-camera>`_
   - Add a Swift file that contains the Main‑Camera code.
   - Add a SwiftUI view that presents the Main Camera output.

2. **Have Vision Pro read OpenAI responses aloud**

   - Use **AVSpeechSynthesizer** to speak the assistant’s reply.

3. **Change the OpenAI model**

   - In *Discover* ▸ `Views/OpenAIService` replace the string
     ``"gpt-4o"`` with the desired model name.

4. **Use a different AI provider**

   - Possible alternatives: **Claude**, **DeepSeek**, **Gemini**.

5. **Combine voice + image pipelines and integrate with PuppyPy**

   - Ensure the two features work together before linking them to PuppyPy.

Back‑End Modifications / Extensions
-----------------------------------

1. **Replace the image‑classification model**

   - File: `data-processing-main/server2.py`  
     Current model: ``"google/vit-base-patch16-224"``
   - Pick a `Hugging Face image‑classification model <https://huggingface.co/models?sort=trending>`_
   - Change the string and restart the server.

Dependencies
------------------------------------

- To add the **Lottie** library in visionOS, follow the instruction below:

     .. image:: /_static/gifs/Add_Dependency.gif
        :width: 800
        :alt: Add Dependency

- Backend packages are listed in
  `data-processing-main/requirements.txt` and `requirements2.txt`.

Backlog
------------------------------------

The backlog is on **GitHub Projects** → `*CS 495 – Apple Vision Pro* <https://github.com/orgs/Apple-Vision-Pro-UA-CS495/projects/1>`_

Style Expectations
------------------------------------

- Can use animations that ship with the **Lottie** dependency.
- Follow the View-Model-ViewModel(MVVM) architecture.

Unit Testing
------------------------------------

#. visionOS

   - visionOS currently does not support ``UI Testing Bundle``, it only has ``Unit Testing Bundle`` so we can only perform logical testing only.

   - To run test cases, click on the diamond icon alongside the function declaration in the editor gutter as below

     .. image:: /_static/gifs/OpenAITest.gif
        :width: 800
        :alt: OpenAI Connection Test


#. Backend

.. code-block:: console

   # activate venv
   source .venv/Scripts/activate
   # run tests
   pytest test.py -v