CYPHER / NEO4J：计算和删除模式理解列表结果中的重复项

Question

I am trying to combine count and distinct with comprehension list and keep getting errors.

I think this explains what I'd like to do, though I'm aware it does not work:

Match (n) WHERE n.id={id}
RETURN n {.id, ems: [(n)<-[:LOTS_OF_PATHS*]-()<-[:EVEN_MORE_PATHS*]-(m) | id:     distinct m.id, count: count(distinct m)]}

Have also tried with some collects in there

Match (n) WHERE n.id={id}
RETURN n {.id, ems: [(n)<-[:LOTS_OF_PATHS*]-()<-[:EVEN_MORE_PATHS*]-(m) | collect(distinct m {id:m.id}), count: count(distinct m)]}

I'm actually collecting the comprehension lists, and ideally would like no repeats within each array, and ideally no repeats amongst the sum of collects.

Match (n) WHERE n.id={id}
RETURN n {.id, collect([(n)<-[:LOTS_OF_PATHS*]-()<-[:EVEN_MORE_PATHS*]-(m) | distinct m.id])+ collect([(n)<-[:OTHER_PATHS*]-()<-[:MORE_PATHS*]-(o) | distinct o.id]) as distinct_ems_and_os} 

The count is less important as I can count the array, but would love to understand it a bit better.

Answer 1

This query (which does not use the fairly limited pattern comprehension) might work for you:

MATCH (n)<-[:LOTS_OF_PATHS*]-()<-[:EVEN_MORE_PATHS*]-(m)
WHERE n.id = $id
WITH COLLECT(DISTINCT m.id) AS ems
RETURN {id: $id, ems: ems, count: SIZE(ems)}

NOTE: variable-length relationships have exponential complexity. You may need to set a reasonable upper bound on the number of hops to avoid running out of memory or long execution times.