Add support for datas UNWIND

Issues Dgraph
, , ,
3

Sceat commented :

Hum Custom blocks seems pretty handy, great proposal ! it would also go nicely with unwinds.

I didn’t even noticed that upsert doesn’t support variables as in the javascript client it seems i can set variables on the query and the mutations

Regarding the unwind statement in cypher it allows to take anything iterable and use it in a graph sequence

UNWIND [1, 2, 3] AS num

with this we have the variable num that represent each value of the array but in sequence, so on the first node found the value will be 1 then on the second node it will be 2 etc…

If one wants to upsert an array of users based on their name, the unwind operation allow to write a compact query

upsert {
	// not sure where you put the bloc (your forum link is broken)
	// so let's say we put it there
	var() {
		user_data AS UNWIND [{name: 'Alice', age: 20}, {name: 'Bob', age: 25}]
	}
	query {
		user as var(func: eq(name, user_data.name))
	}
	mutation {
		set {
			uid(user) <age> val(user_data.age) .
		}
	}
}

// the above would set both Alice and Bob with their respective ages

without unwind you actually have to do the following

upsert {
	query($alice: string, $bob: string) {
		alice as var(func: eq(name, $alice))
		bob as var(func: eq(name, $bob))
	}
	mutation($alice_age: int, $bob_age: int) {
		set {
			uid(alice) <age> val($alice_age) .
			uid(bob) <age> val($bob_age) .
		}
	}
}
// it works but in case you have 100 users your generated query is a bit insane