UNION — alternative patterns

{ A } UNION { B } returns the solutions of A followed by the solutions of B. A variable that only one branch binds is null in rows produced by the other branch.

Examples use the sample data.

Example

SELECT * FROM pgrdf.sparql($$
  PREFIX foaf: <http://xmlns.com/foaf/0.1/>
  SELECT ?p ?name ?mbox
  WHERE { { ?p foaf:name ?name }
          UNION
          { ?p foaf:mbox ?mbox } }
$$);
--  {"p": "http://example.org/alice", "mbox": null, "name": "Alice"}
--  {"p": "http://example.org/bob", "mbox": null, "name": "Bob"}
--  {"p": "http://example.org/carol", "mbox": null, "name": "Carol"}
--  {"p": "http://example.org/alice", "mbox": "mailto:alice@example.org", "name": null}
--  {"p": "http://example.org/carol", "mbox": "mailto:carol@example.org", "name": null}

Rows from the first branch carry ?name, rows from the second carry ?mbox. ?p appears in both branches, so every row has it.

Either of two properties

Bind the same variable in both branches to read "this property or that one" into one column:

SELECT * FROM pgrdf.sparql($$
  PREFIX foaf: <http://xmlns.com/foaf/0.1/>
  PREFIX ex:   <http://example.org/>
  SELECT ?who ?contact
  WHERE { { ?who foaf:mbox ?contact }
          UNION
          { ?who ex:phone ?contact } }
  ORDER BY ?who
$$);
--  {"who": "http://example.org/alice", "contact": "mailto:alice@example.org"}
--  {"who": "http://example.org/carol", "contact": "mailto:carol@example.org"}

(Nobody in the sample data has an ex:phone, so only mailboxes come back.)

Combining with other features

• Aggregates and GROUP BY work over a UNION.
• Each branch can be scoped to a graph: { GRAPH <g1> { … } } UNION { GRAPH <g2> { … } } — see GRAPH.
• A UNION inside OPTIONAL or inside MINUS is refused with an error. See OPTIONAL for a workaround.