Single-Hop Neighbors

Where we left off

Lesson 6 fetched exactly one node. This lesson takes the first real graph step: from a node, who (or what) is directly connected to it.

`graph.get_neighbors()`

SELECT node_id, depth, node FROM graph.get_neighbors(
  graph_name := 'default',
  label := 'people',
  id := 'p1',
  direction := 'out',
  edge_types := ARRAY['works_at']
);

This is a single hop only, everything one edge away from p1, not a multi-hop walk (that's graph.traverse(), next lesson). edge_types restricts which relationship types to follow, useful once a graph has more than one (Lesson 10 onward will). Leaving it unset would follow every registered edge type.

`direction`: which way to walk

direction takes exactly three values: 'out', 'in', or 'any', not the more readable-looking 'outgoing'/'incoming' you might guess from the parameter name, pggraph is strict about this and errors if you pass anything else. 'out' follows the edge the way it was declared (people.company_id -> companies.id), so from a person it reaches their company. Because Lesson 4 registered works_at as bidirectional := true, 'in' from a *company* would also reach its people, even though the foreign key itself only points one way.

Reading the result

node_id=c1  depth=1  node={"id": "c1", "name": "Acme Bank"}

depth is always 1 for get_neighbors(), since it's single-hop by definition, it's included because graph.traverse() returns rows in the same shape with depth actually varying.