SPARQL: Fixes DISTINCT after ORDER BY

lib/src/sparql/eval.rs

@@ -295,17 +295,8 @@ impl<S: EncodedQuadsStore> SimpleEvaluator<S> {
             }
             PlanNode::HashDeduplicate { child } => {
                 let iter = self.eval_plan(&*child, from);
-                let mut values = HashSet::with_capacity(iter.size_hint().0);
+                let already_seen = HashSet::with_capacity(iter.size_hint().0);
-                let mut errors = Vec::default();
+                Box::new(HashDeduplicateIterator { iter, already_seen })
-                for result in iter {
-                    match result {
-                        Ok(result) => {
-                            values.insert(result);
-                        }
-                        Err(error) => errors.push(Err(error)),
-                    }
-                }
-                Box::new(errors.into_iter().chain(values.into_iter().map(Ok)))
             }
             PlanNode::Skip { child, count } => Box::new(self.eval_plan(&*child, from).skip(*count)),
             PlanNode::Limit { child, count } => {
@@ -1048,6 +1039,26 @@ impl<'a, S: EncodedQuadsStore> Iterator for UnionIterator<'a, S> {
     }
 }
 
+struct HashDeduplicateIterator<'a> {
+    iter: EncodedTuplesIterator<'a>,
+    already_seen: HashSet<EncodedTuple>,
+}
+
+impl<'a> Iterator for HashDeduplicateIterator<'a> {
+    type Item = Result<EncodedTuple>;
+
+    fn next(&mut self) -> Option<Result<EncodedTuple>> {
+        match self.iter.next()? {
+            Ok(tuple) => if self.already_seen.insert(tuple.clone()) {
+                Some(Ok(tuple))
+            } else {
+                self.next()
+            },
+            Err(error) => Some(Err(error)),
+        }
+    }
+}
+
 struct ConstructIterator<'a, S: EncodedQuadsStore> {
     store: Arc<S>,
     iter: EncodedTuplesIterator<'a>,

lib/tests/sparql_test_cases.rs

@@ -82,6 +82,7 @@ fn sparql_w3c_syntax_testsuite() {
 
 #[test]
 fn sparql_w3c_query_evaluation_testsuite() {
+    //TODO: dataset graph open-world regex
     let manifest_10_urls = vec![
         Url::parse("http://www.w3.org/2001/sw/DataAccess/tests/data-r2/algebra/manifest.ttl")
             .unwrap(),
@@ -115,6 +116,10 @@ fn sparql_w3c_query_evaluation_testsuite() {
         ).unwrap(),
         Url::parse("http://www.w3.org/2001/sw/DataAccess/tests/data-r2/optional/manifest.ttl")
             .unwrap(),
+        Url::parse("http://www.w3.org/2001/sw/DataAccess/tests/data-r2/reduced/manifest.ttl")
+            .unwrap(),
+        Url::parse("http://www.w3.org/2001/sw/DataAccess/tests/data-r2/solution-seq/manifest.ttl")
+            .unwrap(),
         Url::parse("http://www.w3.org/2001/sw/DataAccess/tests/data-r2/sort/manifest.ttl").unwrap(),
         Url::parse("http://www.w3.org/2001/sw/DataAccess/tests/data-r2/triple-match/manifest.ttl")
             .unwrap(),