wip: Demo of using pagefind search

[corneliusroemer]

relates to #3079

preview URL: https://search.loculus.org

Summary

I've copied a dummy index over from pathoplexus, to show how it would work. So if you click on links, most go 404 - that's expected.

The CSS is still super ugly, needs some care to port over from starlight (or adaptation/redesign, but I prefer the modal over the pagefind default UI)

Ignore the many index files in /public, this is just to make the preview work

Screenshot

[theosanderson]

This is cool!

I think seeing a search box at the top of a database I might have an initial expectation that it would search the database rather than (just) the docs. I guess one thing we could do would be to detect accession-like search entries and make a special link for them.

[corneliusroemer]

I was mostly thinking about docs search as the use case here, e.g. locating the minutes - Pagefind might not scale to hundreds of thousands of pages. But indeed, we could intercept searches for things like PP_ or similar and don't give them to Pagefind but use our own search engine of kinds.

Would indeed be cool if there was a global search one could just paste an accession into and get results - hadn't thought of that at all until you mentioned it.

[theosanderson]

Yeah basically I think a docs-only search would ideally have a different presentation to make clear that that's what it is.

The search box here

would confuse me if it searched docs :)

[theosanderson]

It would be cool to include the top level organism pages like https://search.loculus.org/cchf/search? too (I know this is just a demo :) )

[corneliusroemer]

Ah yeah - it's easy to add extra pages to the index. I created the index as explained here: #3079 (comment)

I'm not sure why the organism search pages weren't indexed, maybe because they are behind a redirect and wget didn't get there as a result. I was just hacking things together, didn't spend much time on any part of it.