Chloe Kudryavtsev
538e81db56
The query is split up on spaces, and we search for each of those terms, in order, anywhere in the emoji name or any aliases/keywords. This is done in a single filter pass against a compiled regex, making the process reasonably performant. Based on rough estimates, it should be between 2 and 5x faster than the old implementation, depending on several factors. There is a natural space left in to sort by relevancy (not done yet). It should also be easy to make the number of matches shown configurable. The number of matches is relevant, especially pre-sort. Another consideration is to delay the calculation by up to 300ms. |
||
---|---|---|
.. | ||
backend | ||
client | ||
foundkey-js | ||
shared | ||
sw | ||
meta.json |