From 258ac21500d133cb26460354e5f0b0063e6b1648 Mon Sep 17 00:00:00 2001
From: Pierre-Antoine Champin <pierre-antoine@w3.org>
Date: Fri, 22 Nov 2024 16:19:10 +0100
Subject: [PATCH] implement SPARQL Union

---
 sparql/src/exec.rs | 27 ++++++++++++++++++++++++++-
 sparql/src/test.rs | 38 ++++++++++++++++++++++++++++++++++++++
 2 files changed, 64 insertions(+), 1 deletion(-)

diff --git a/sparql/src/exec.rs b/sparql/src/exec.rs
index e6fc22c..1f5314c 100644
--- a/sparql/src/exec.rs
+++ b/sparql/src/exec.rs
@@ -101,7 +101,7 @@ impl<'a, D: Dataset + ?Sized> ExecState<'a, D> {
                 expression,
             } => Err(SparqlWrapperError::NotImplemented("LeftJoin")),
             Filter { expr, inner } => self.filter(expr, inner, graph_matcher, binding),
-            Union { left, right } => Err(SparqlWrapperError::NotImplemented("Union")),
+            Union { left, right } => self.union(left, right, graph_matcher, binding),
             Graph { name, inner } => Err(SparqlWrapperError::NotImplemented("Graph")),
             Extend {
                 inner,
@@ -184,6 +184,31 @@ impl<'a, D: Dataset + ?Sized> ExecState<'a, D> {
         Ok(Bindings { variables, iter })
     }
 
+    fn union(
+        &mut self,
+        left: &GraphPattern,
+        right: &GraphPattern,
+        graph_matcher: &[Option<ArcTerm>],
+        binding: Option<&Binding>,
+    ) -> Result<Bindings<'a, D>, SparqlWrapperError<D::Error>> {
+        let Bindings {
+            variables: lv,
+            iter: li,
+        } = self.select(left, graph_matcher, binding)?;
+        let Bindings {
+            variables: rv,
+            iter: ri,
+        } = self.select(right, graph_matcher, binding)?;
+        let mut variables = lv.clone();
+        for v in rv {
+            if lv.iter().all(|i| *i != v) {
+                variables.push(v)
+            }
+        }
+        let iter = Box::new(li.chain(ri));
+        Ok(Bindings { variables, iter })
+    }
+
     fn extend(
         &mut self,
         inner: &GraphPattern,
diff --git a/sparql/src/test.rs b/sparql/src/test.rs
index e9dda63..73457ff 100644
--- a/sparql/src/test.rs
+++ b/sparql/src/test.rs
@@ -19,6 +19,11 @@ use test_case::test_case;
     vec![];
     "no result"
 )]
+#[test_case(
+    "PREFIX s: <http://schema.org/>  SELECT ?x { { ?x a s:Event } UNION { ?x a s:Person } }",
+    vec!["<https://example.org/test#a>", "_:b"];
+    "union"
+)]
 fn test_select_1_and_ask(query: &str, exp: Vec<&str>) -> TestResult {
     let dataset = dataset_101()?;
     let dataset = SparqlWrapper(&dataset);
@@ -35,6 +40,39 @@ fn test_select_1_and_ask(query: &str, exp: Vec<&str>) -> TestResult {
     Ok(())
 }
 
+#[test]
+fn test_union() -> TestResult {
+    let dataset = dataset_101()?;
+    let dataset = SparqlWrapper(&dataset);
+    let parsed_query = SparqlQuery::parse(
+        r"
+        PREFIX s: <http://schema.org/>
+        SELECT ?p ?e {
+            { ?p a s:Person }
+            UNION
+            { ?e a s:Event }
+        }
+    ",
+    )?;
+    let bindings = dataset.query(&parsed_query)?.into_bindings();
+    assert_eq!(bindings.variables(), &["p", "e"]);
+    let mut got = bindings.into_iter().collect::<Result<Vec<_>, _>>()?;
+    got.sort();
+    assert_eq!(got.len(), 2);
+    assert_eq!(got[0].len(), 2);
+    assert!(got[0][0].is_none());
+    assert!(got[0][1].is_some());
+    assert!(got[0][1].as_ref().unwrap().is_blank_node());
+    assert!(got[1][0].is_some());
+    assert!(got[1][0].as_ref().unwrap().is_iri());
+    assert_eq!(
+        got[1][0].as_ref().unwrap().to_string(),
+        "<https://example.org/test#a>"
+    );
+    assert!(got[1][1].is_none());
+    Ok(())
+}
+
 #[test_case(1)]
 #[test_case(2)]
 #[test_case(3)]