fix: #3491 - Unable to use tensorrt-llm

[louis-menlo]

Describe Your Changes

This PR fixes the Tensor-LLM engine download URL, making it accessible to Windows users with supported Nvidia GPUs, who can now install and use Tensort-LLM models.

This also fixes the incorrect supported platforms from the extension, which currently only support Windows.

There's an edge case with model downloading on Windows that's been fixed, and it also resolved an issue where the renderer process would send a wrong aborting ID.

Screenshots

Install Tensorrt-llm engine on Windows

Fixes Issues

• bug: Fix Download location of Nitro TensorRT on Windows #3491

1. In package.json:

   • Added Jest testing setup and dependencies
   • Updated the scripts section to include a test command

2. In rollup.config.ts:

   • Updated the download URLs for the runner, changing from janhq/nitro-tensorrt-llm to janhq/cortex.tensorrt-llm

3. In src/index.ts:

   • Removed the supportedPlatform array
   • Updated the compatibility check to use this.compatibility().platform instead of this.supportedPlatform

4. In tsconfig.json:

   • Added an exclude field to ignore test files

5. In web/hooks/useDownloadModel.ts:

   • Updated the abortModelDownload function to handle ModelFile instead of Model
   • Modified the path construction logic in abortModelDownload to use file_path when available

These changes mainly focus on setting up Jest for testing, updating download URLs, refactoring some compatibility checks, and adjusting the model download abort functionality.

[louis-menlo]

Hmmm..checking

[github-actions]

Barecheck - Code coverage report

Total: 64.8%

Your code coverage diff: 0.00% ▴

Uncovered files and lines

File	Lines
web/hooks/useDownloadModel.ts	41-42, 60, 77-79, 97, 99, 102, 118