Reputation: 943
I find weird that using OPTIONAL MATCH nodes that don’t have the expected relationship are not returned as a single node in path.
OPTIONAL MATCH path = (:Person) -[:LIKES]- (:Movie)
UNWIND nodes(p) as n
UNWIND rels(p) as e
WITH n
WHERE HEAD(LABELS(n)) = “Person”
return COUNT(DISTINCT n)
The number of people returned only includes those who liked a movie. By using OPTIONAL I would have expected all people to be returned. Is there a workaround to this or am I doing some this wrong in the query?
Upvotes: 0
Views: 168
Reputation: 30407
A better way to go about this would be to match to all :People nodes first, then use the OPTIONAL MATCH to match to movies (or, if you want a collection of the movies they liked, use pattern comprehension).
If you do need to perform an UNWIND on an empty collection without wiping out the row, use a CASE around some condition to use a single-element list rather than the empty list.
MATCH (n:Person) // match all persons
OPTIONAL MATCH p = (n) -[:LIKES]- (m:Movie) // p and m are the optionals
UNWIND CASE WHEN p is null THEN [null] ELSE nodes(p) END as nodes // already have n, using a different variable
UNWIND CASE WHEN p is null THEN [null] ELSE rels(p) END as e // forcing a single element list means UNWIND won't wipe out the row
WITH n
WHERE HEAD(LABELS(n)) = “Person” // not really needed at all, and bad practice, you don't know the order of the labels on a node
return COUNT(DISTINCT n) // if this is really all you need, just keep the first match and the return of the query (without distinct), don't need anything else
Upvotes: 1