We would need to mark translators that require running via a headless browser. Some only need to wait for JS on search pages, which is not important for TS.
A headless browser would be spun up via puppeteer for those translators and wait some set time before running translation.
An investigation into resource usage is required
An investigation into sandboxing cookies/localstorage is required between translation runs
An investigation into running on AWS lambda is required
adomasven
commented
3 weeks ago
Translation solely via XHR is an issue on some translators (currently 70+ translators have test cases with
defer: true), primarily because additional loading via JS happens on those websites (e.g. https://www.fao.org/documents/card/en?details=cc0461en)