Reputation: 155
Lets say we have a Neo4j graph such as (Brand)-[:from]->(Post)<-[:likes]-(Person).
How can I return a cypher query which will have a minimum number of brand posts, say 3. I want this to be scalable and not dependent on a specific property attribute value.
Hence the results would return at least 3 instances of the Brand nodes, as well as maybe 5 from Post and 15 from Person.
I have tried a few different things:
1.) Declare several variable names for each brand (not scalable)
Match (b:Brand)-[]->(p:Post)<-[]-(per:Person)
Match (b1:Brand)-[]->(p1:Post)<-[]-(per2:Person)
Match (b2:Brand)-[]->(p2:Post)<-[]-(per3:Person)
return b,b1,b2,p,p1,p2,per,per2,per3
limit 30
This didn't work because it essentially return the same as
Match (b:Brand)-[]->(p:Post)<-[]-(per:Person)
return b,p,per
limit 30
2.) Use a foreach some
Match (b:Brand) WITH collect (distinct b) as bb
FOREACH (b in bb[0..3] | MATCH (b)-[]->(p:Post)<-[]-(per:Person))
RETURN b, p, per LIMIT 40
This didn't work because you can't use Match inside a Foreach call.
The only way I know how to do this is to declare a where clause with their unique property brand name values which is not scalable. It looks like this:
Match (b:Brand)-[]->(p:Post)<-[]-(per:Person)
where b.brand = "b1" OR b.brand ="b2" or b.brand = "b3"
Return b,p,per
Limit 30
However the above still doesn't even return what I want.
Please help. Here is a quick graph to test on:
Create (b1:Brand {brand:'b1'})
Create (b2:Brand {brand:'b2'})
Create (b3:Brand {brand:'b3'})
Create (p1:Post {id: "001",message: "foo"})
Create (p2:Post {id: "002",message: "bar"})
Create (p3:Post {id: "003",message: "baz"})
Create (p4:Post {id: "004",message: "raz"})
Create (per1:Person {id: "001",name: "foo"})
Create (per2:Person {id: "002",name: "foo"})
Create (per3:Person {id: "003",name: "foo"})
Create (per4:Person {id: "004",name: "foo"})
Create (per5:Person {id: "005",name: "foo"})
Create (per6:Person {id: "006",name: "foo"})
Create (per7:Person {id: "007",name: "foo"})
Merge (b1)-[:FROM]->(p1)
Merge (b1)-[:FROM]->(p2)
Merge (b2)-[:FROM]->(p3)
Merge (b3)-[:FROM]->(p4)
Merge (per1)-[:LIKES]->(p1)
Merge (per1)-[:LIKES]->(p2)
Merge (per1)-[:LIKES]->(p3)
Merge (per2)-[:LIKES]->(p1)
Merge (per2)-[:LIKES]->(p4)
Merge (per3)-[:LIKES]->(p3)
Merge (per4)-[:LIKES]->(p1)
Merge (per5)-[:LIKES]->(p2)
Merge (per6)-[:LIKES]->(p1)
Merge (per6)-[:LIKES]->(p2)
Merge (per6)-[:LIKES]->(p3)
Merge (per6)-[:LIKES]->(p4)
Merge (per7)-[:LIKES]->(p4)
Upvotes: 1
Views: 475
Reputation: 29172
You can use the unwind
instead of foreach
:
Match (b:Brand) WITH collect (distinct b) as bb
UNWIND bb[0..3] as b
MATCH (b)-[]->(p:Post)<-[]-(per:Person)
RETURN b, p, per LIMIT 40
Or combine with
and limit
:
MATCH (b:Brand) WITH distinct b LIMIT 3
MATCH (b)-[]->(p:Post)<-[]-(per:Person)
RETURN b, p, per LIMIT 40
Upvotes: 0