Performance issue with large list and long search text

I’m not sure am I doing something wrong, but I noticed that when doing search with a very long string and quite low threshold of 0.3, the search becomes very slow ~8-900ms with MBP 2015.

It’s hard for me to pinpoint exactly the cause, but since there isn’t issues searching with a short string, I guess the longer string just isn’t handled well by a computation, that isn’t exactly O(1).

https://imgur.com/a/T9OYGJ1

That’s the lowest level where my Chrome JS profiler was able to get to. Doing simple performance check with:

      console.time('search')
      const newResults = fuse.search(newVal)
      console.timeEnd('search')

confirms that this indeed is the culprit.

A value for newVal might be eg Ajoneuvot ja liikenne / Autot / Automerkit / Volkswagen / Bora, Vento ja Jett. Fuse is provided a list, where the first searched value is string (similar to that previous one), the second the splitted version of that same string from ‘/’ eg ['Ajoneuvot ja liikenne', 'Autot', 'Automerkit', 'Volkswagen', 'Bora, Vento ja Jett']. The options object being: {threshold: 0.3, keys: ['text', 'breadcrumb'], maxPatternLength: 32 }.

My search items list is 2427 long, with a single instance being eg:

  {
    breadcrumb: ["Ajanviete"]
    key: 52
    text: "Ajanviete"
    value: 52
  }

[TeemuKoivisto]

I made a demo here https://teemukoivisto.github.io/search-demo-app/

Quite weirdly, it is much faster when I have Chrome’s developer tools hidden. Now it’s even usable on my Android. But it’s still laggy and the worst of all, when I have a specific topic chosen, topics close to it don’t show in sequential fashion.

Eg I have topic Harrastukset / Pelit / Vedonlyönti ja rahapelit / Pokeri / Pokeriturnaukset chosen, yet after first 10 or so items I start getting topics from way off the original eg Viihde ja kulttuuri / Kirjallisuus / Sarjakuvat / Omat sarjakuvat. It should be showing topics from the same parent topic eg Harrastukset / Pelit / Vedonlyönti ja rahapelit / Yleistä vedonlyönnistä and work its way down. I added the tokenize: true parameter but it didn’t help.

I guess this means I have to make my own custom search. I had hoped I would avoid using dull substring search but I guess I have to make use of it. Maybe use only fuzzy search for the last 10 characters of the string.

[peteroid]

I am facing the same issue as on my React Native app. There are 2k items and the performance significantly dropped when the search text was growing from 1 to 100 chars.

Setting the threshold to lower value (lower fuzziness) would help a bit.