How nested are the fields you are
How nested are the fields you are searching inside the object? one possible approach might be to normalize your data more
7 Replies
mainPhoneNumber: v.string(),
if i have to search this i can do
q = q.filter((q) => q.eq(q.field("mainPhoneNumber"), args.phone));
lets say this
phoneNumbers: v.array(v.object({type: v.string(),number: v.string(),})),
now in my head i can do q.eleMatch
or just try filter with phoneNumbers.number
which in case of trpc works.
but not in convex.
i didn't explore more q. functions though.
if you know, i'm happy to listen carefully! 🙂there's an equally-efficient alternative to
.filter which allows you to do this (which is to say, neither are efficient and indexes are more efficient) https://stack.convex.dev/complex-filters-in-convexStack
Using TypeScript to Write Complex Query Filters
There’s a new Convex helper to perform generic TypeScript filters, with the same performance as built-in Convex filters, and unlimited potential.
ah, thanks lee. will you still suggestion in-memory filtering in convex ts file though?
because this works perfectly for me , but obviously i'm fetching all clients from database here. (i can have filters to cut that down though
)
const clients = await ctx.db.query("clients").collect();
const phoneNumberExists = clients.some(client =>
client.phoneNumbers.some(phone => phone.number === phoneNumber)
);
as long as you don't have many clients, this looks great to me. if you have many clients, filters won't help; you need indexes
gotcha! i am obviously not going to hurt the db to get all clients. there will be thousands. i'll cut the initial query down
and ofcourse, i'm on the line to do indexes because i also want to explore the vector and db search later!
thanks
great! The doc mentioned above describes how to do the filter you want using indexes, btw
return await filter(clientsQuery, (client) =>client.mainPhoneNumber === args.phone || client.phoneNumbers.some(p => p.number === args.phone)).collect();
beautifully works!
i'll be handling indexes seperately later.
thanks again!