Clarification on multiple MATCH patterns in a Cypher query

Question

In the below query, does the 2nd match pattern john-[r?:HAS_SEEN]->(movie) run on the result of the first match john-[:IS_FRIEND_OF]->(user)-[:HAS_SEEN]->(movie) . I am trying to understand if this is similar to the unix pipe concept i.e. the result of the 1st pattern is the input to the 2nd pattern.

start john=node(1)
match
john-[:IS_FRIEND_OF]->(user)-[:HAS_SEEN]->(movie),
john-[r?:HAS_SEEN]->(movie)
where r is null
return movie;

Answer 1

I don't think I would compare multiple MATCH clauses to the UNIX pipes concept. Using multiple, comma-separated matches is just a way of breaking out of the 1-dimensional constraint of writing relationships with a single sentence. For example, the following is completely valid:

MATCH a--b, 
      b--c, 
      c--d, 
      d--e,
      a--c

At the very end I went back and referenced a and c even though they weren't used in the clause directly before. Again, this is just a way of drawing 2 dimensions' worth of relationships by only using 1-dimensional sentences. We're drawing a 2-dimensional picture with several 1-dimensional pieces.

On a side note, I WOULD compare the WITH clause to UNIX pipes -- I'd call them analogous. WITH will pipe out any results it finds into the next set of clauses you give it.

Answer 2

Yes, simply think of these two matches to be one - i.e.

match (movie)<-[r?:HAS_SEEN]-john-[:IS_FRIEND_OF]->(user)-[:HAS_SEEN]->(movie)

or

match john-[:IS_FRIEND_OF]->(user)-[:HAS_SEEN]->(movie)<-[r?:HAS_SEEN]-john