Groupby + Orderasc in Query

Users Dgraph
1

What I want to do

Return the most common Field.value

What I did

This returns the fields with a “count” field, but I’m not sure how to then order and limit to 1:

{
  query(func: uid("0xc44c")) {
    Organization.fields @groupby(Field.value) {
    	count(uid)
    }
  }
}

This returns the error: Vars can be assigned only when grouped by UID attribute

{   
  var(func: uid("0xc44c")) {
    Organization.fields @groupby(Field.value) {
    	a as count(uid)
    }
  }
  
  query(func: uid(a)) { uid }
}
2

Groupy by works only on edges, not values.

3

To clarify, Field is a node type with an edge called Field.value :slight_smile:

4 
{   
  query(func: uid("0xc44c")) {
		uid
    Organization.fields @groupby(Field.value) {
    	count(uid)
    }
  }
}

Returns:

{
  "data": {
    "query": [
      {
        "uid": "0xc44c",
        "Organization.fields": [
          {
            "@groupby": [
              {
                "Field.value": "https://www.google.com/",
                "count": 1
              },
              {
                "Field.value": "https://www.nytco.com/",
                "count": 2
              }
            ]
          }
        ]
      }
    ]
  }

but i’m unsure how to then order by “count”

5

You can’t, cuz Group By works only on edges. And I know that is possible to run the first step on value edges. But it is not foreseen.

6

Sorry, I’m not following what you mean.

In my example above I’m grouping by an edge called “Field.value”, and the query returns the strings these edges are pointing to along with a “count” of how common each string is.

Are you saying there’s no way to then sort by this count?

7

Yes.

Count with Indexing is a post computed feature. So you need an extra block to do so. But you can’t do it with GB cuz it wasn’t made for value edges.

8

ah, if it’s not too much trouble could you please show me what you mean by this?

when i try to create another block, I get this error: “Vars can be assigned only when grouped by UID attribute”

9

In fact the phrase would be ‘So would you need an extra block to do so.’

That’s what I’m trying to say. The Group By is a multi-block feature. Group By doesn’t work with Value Edges. Only relational edges. That’s why you get this error.

10

To take a step back and try and understand things a bit better, if my the schema is:

type Form {
    id: ID!
    fields: [uid]
}
type Field {
    id: ID!
    name: string
    value: string
}

How would I query the most common value for each field name? is there an approach other than groupby I could use?

11

@MichelDiz On the example on this page it’s my understanding that genre is a value edge, am I wrong?

https://dgraph.io/docs/query-language/groupby/#user

{
  var(func:allofterms(name@en, "steven spielberg")) {
    director.film @groupby(genre) {
      a as count(uid)
      # a is a genre UID to count value variable
    }
  }

  byGenre(func: uid(a), orderdesc: val(a)) {
    name@en
    total_movies : val(a)
  }
}

When I try this query

{
  var(func: eq(Node.name, "Example"))  {
    uid 
    Node.references @groupby(Reference.description) {
         a as count(uid)
    }
  }
  
  query(func: uid(a), orderdesc: val(a)) {
    description: val(a)
  }
}

I get: Vars can be assigned only when grouped by UID attribute

12

Nope, the genre is a relational edge. It goes from “film” to “genre” type. And the groupby captures the value from the count and does the grouping.

This feels like a value edge, it won’t work.

That is the error I mentioned why. Group By works only in relational edges. We could add this feature, but I don’t see this happening soon.

Cheers.