Wildcard query on Firebase?

Question

Is it possible to do wildcard queries on Firebase? For example:

https://foo.firebaseio.com/person.json?orderBy="name"&equalTo="Lun*"

Solution 1

I know it's been a while but I thought that others might be interested. You can "fake" a wildcard search for things like foo* (so basically you can search for values beginning with a specified string).

For iOS & Swift it would look like this:

dbReference.child("person").queryOrdered(byChild: "name").queryStarting(atValue: "foo").queryEnding(atValue: "foo\u{f8ff}").observe(.childAdded) { (snapshot: FIRDataSnapshot) in
    print("\(snapshot.key) - \(String(describing: snapshot.value))")
}

What this does is using a start and end values for name property where the end key is equal to the start + a very high code point in the Unicode range. Because it is after most regular characters in Unicode, the query matches all values that start with foo.

Solution 2

No. But kinda.

You cannot do a wildcard query, however, you can structure your data that will allow for this.

For example, say we want to find matches for users whose name starts with Ler

Here's our structure

users
  uid_0
    name: "Leroy"

Store the decomposed data in another node: Remember, disk space is cheap.

decomposed
  uid_0
    L: true
    Le: true
    Ler: true
    Lero: true
    Leroy: true

then perform a query on the decomposed node for the value of true for children equal to Ler

ref.queryOrderedByChild("Ler").queryEqualToValue(true).observeEventType(.ChildAdded, 
     withBlock: { snapshot in
                    print(snapshot.key)
                })

And the snapshot.key will be uid_0

Comments

[MoreScratch]

That's a interesting way to achieve what I need. Only issue is that I need to do a wildcard for any part of a name (e.g. roy). Also, I am going to have ~200K people in my db. Will it scale?

[Jay]

It can scale, (Firebase is scalable) however, it's probably not realistic for that size a dataset. If you were just looking for something like an autofill, where you type the first few chars of say, a last name, then it's more realistic. We crafted a ObjC class that we can throw a name at and it will decompose the name into all searchable combinations and then it stores that in a Firebase node; with that it feels pretty automated, but we haven't tried working with thousands of names. It's probably a good question for the Firebase team.

[Oswaldo]

Imagine you want to do a wildcard search in all string fields of your database, it's impracticable to decompose each value of each field in N combinations

[Jay]

@Oswaldo As I posted in my prior comment, it's applicable in some situations where, for example you have one field you want to query out of 100 users. Decomposing the string 'feels' like a lot of data but if it's a short string, it's only a couple hundred bytes times 100 or even 1000 users. That's a tiny amount of storage but it is highly situational. And thanks for the down vote even through I made it clear it's not for large datasets. We use it all the time and it works VERY well if used properly.

[Oswaldo]

@Jay, I downvoted because the quetion was more general use case and not specific like yout answer. However I found this question very usefull for me and the community and would like to have a more satisfactory answer, dont deserves 2 upvotes for a very specific use case answer

[mamu]

how is that an answer for your question?

[MoreScratch]

I needed to close the loop on the question. It was clear to me that a solution does not exist within Firebase so we decided to simply change our BaaS provider. In essence there is no answer to be had that we can see based on the documentation and the responses to the question.

[Learn2Code]

You offer no explanation of how it solved your problem, or did it really answer your question.

[MoreScratch]

It solved my problem because I decided to go with another solution that supported the functionality that I required. I thought that was clear. In any case I un-marked it as the answer because it has proven to be confusing for some people.

[cjbarth]

This is a great solution. If you want to cover emoji and other characters too, you can use the largest permissible Unicode character: \u{10ffff}.

[Hesam]

For Android something like this: firestore?.collection("YOUR_COLLECTION") ?.orderBy("YOUR_KEY", Query.Direction.ASCENDING) ?.startAt(tag) ?.endAt("$tag\\uf8ff") ?.limit(10) ?.get() ?.addOnSuccessListener { ... }