Change EmojiTrie implementation to SuffixTree

[hannojg]

For this issue ticket we have implemented a very efficient/fast suffix tree implementation:

• [HOLD for payment 2025-01-07] [Performance] Typing in search is very laggy for big accounts #46591

PR:

• Search suffix tree implementation #48652

It was discussed that we want to use the implementation of the SuffixTree for the emoji search as well:

• Discussion here

This enables:

• more efficient search in emojis
• faster construction of the emoji tree (we should compare the construction time before and after the change.)
• searching for any substring of an emoji name (not just perfect name matches)

[hannojg]

cc @roryabraham

[ChavdaSachin]

Hey @hannojg, I have been very passionate Competitive Programmer back during my college days.
I think I could be a help here.
Do you think there is a scope for an External contributor here?

[hannojg]

I think this is more of an internal ticket. The suffix tree implementation has already been built by our internal team and this is a follow up issue from one of our PRs.

[ChavdaSachin]

Alright,
But I have one more suggestion here, we could make use of c++ here instead of js , c++ is well known for it's speed, I am really positive that c++ would make significant improvement here.

[melvin-bot]

@hannojg, this Monthly task hasn't been acted upon in 6 weeks; closing.

If you disagree, feel encouraged to reopen it -- but pick your least important issue to close instead.

[hannojg]

@mountiny can you please reopen? The initial PR for changing the search algorithm to a search tree took longer than anticipated. We are going to look into these items next, so its still l a priority!

[hannojg]

But I have one more suggestion here, we could make use of c++ here instead of js , c++ is well known for it's speed, I am really positive that c++ would make significant improvement here.

Yes, while c++ might be faster there is a significant cost for bringing the data from JS over to C++ which might eat away the performance benefits. The current typescript implementation is already pretty fast.
We could open a follow up ticket to see if there is any benefit in bringing this to C++, but for now we are good I'd say