Ref: CREATE - Neo4j Cypher Manual
MATCH (a:Person),(b:Person)
WHERE a.name = 'A' AND b.name = 'B'
CREATE (a)-[r:RELTYPE { name: a.name + '<->' + b.name }]->(b)
RETURN type(r), r.name
This above is how Neo4J deals with their “Facet” / relationship proprieties. And in this case, the query makes a “concat” between the values of both nodes, on the same edge.
in Graphql +- would look like.
upsert {
query {
A as var(func: type(Person)) @filter(eq(name, "A")) { a as name }
B as var(func: type(Person)) @filter(eq(name, "B")) { b as name }
}
mutation {
set {
uid(A) <RELTYPE> uid(B) (name="val(a)+'<->'val(b)").
}
}
}
In this case, we need a “concatenate” inside the double quotes for values and pass the val properties.
Implicit vals
Also another ref: Improve Facets mutation handling · Issue #1996 · dgraph-io/dgraph · GitHub
If we have @facets(r as rating ...) in theory variables can catch the predicate name and the value recursively. So for Facets this would be nice. Cuz when it expands the val it will expand the attribute and value when using the variable in the mutation.
So instead of val(r)=val(r) we just val(r).
upsert {
query {
var(func: eq(email, "aman@dgraph.io")) {
v as uid
edge as someEdge @facets(r as rating, t as something, e as somethingElse)
}
}
mutation {
set {
uid(v) <someEdge> uid(edge)
(weight=0.1, updated_at=2019-06-02T13:01:09, val(r), val(t), val(e)) .
#OR
(weight=0.1, updated_at=2019-06-02T13:01:09, rating=val(r), something=val(t), somethingElse=val(e)) .
}
}
}
I vote to always simplify it. The smaller and more direct the query is, the better.
Recurse value variable
upsert {
query {
var(func: eq(email, "aman@dgraph.io")) {
v as uid
edge as someEdge @facets(r as ...)
}
}
mutation {
set {
uid(v) <someEdge> uid(edge) (
val(r) # In this variable will enter all data we had previously. Recursively.
weight=0.1,
updated_at="2019-06-02T13:01:09",
) .
}
}
}
The three dots would be the operator (r as ...). It will copy and paste the previous values together with the new values.
In this case, “weight” and “updated_at” already exist and will be captured by the “Recursive Variable”. However, at the time of mutation, the new data will prevail.