Create JS Context for PDF worker

[mvasilak]

Creates a JS Context to load the PDF worker in, that will run solely in a background queue.

[mvasilak]

@mrtcode you can check in this PR how the PDF worker is used in a JSContext, via this js shim. Previously we used this html shim which we loaded in the web view. Any feedback is welcome.

[mrtcode]

@mvasilak great, thanks! I'll try it soon. I'll need to setup Xcode…

[mrtcode]

I now have an Xcode environment set up and can build the app, but I’m not sure what exactly I should be testing. Is this already expected to work, or does it currently just load the worker.js code via the shim?

What I really need is a test in the pdf-worker repo that runs the worker in a JSContext environment. I briefly tried to vibe code some Swift code that loads the shim and executes worker.js. At first glance, it evaluates without errors, although I haven’t yet tried loading an actual PDF file.

[mvasilak]

@mrtcode it is expected to work. In app you can test it with retrieve metadata, or if you enable the speech feature, which (for the time being) uses the PDF worker to extract text to read. Furthermore, there are tests that do test PDF worker metadata and text extraction. If you have more tests of your own, you could vibe convert them to the iOS test harness.

All in all, just taking a look at the JS shim file and commenting on that if needed, should suffice, as this is the vibe coded layer that glues the PDF worker to the JSContext environment.

[mrtcode]

it is expected to work

Great job then!

So there are many changes coming to pdf-worker (document-worker), such as model inference, and I’ll set up JSContext tests in its repo to make sure everything continues to work.

[mvasilak]

@michalrentka ready for another quick look, updated to properly use the latest pdf-worker, with some extra tests.

[mrtcode]

Note that I also added JSContext tests to the new pdf-worker (which will be renamed to document-worker) https://github.com/zotero/pdf-worker/tree/document-worker/test/jscontext