Fix assertion error for GRAPH with unconnected components

ad-freiburg · Feb 28, 2025 · 21c2e61 · 21c2e61
1 parent 8fe0642
commit 21c2e61
Show file tree

Hide file tree

Showing 3 changed files with 73 additions and 3 deletions.
diff --git a/src/engine/QueryPlanner.cpp b/src/engine/QueryPlanner.cpp
@@ -486,15 +486,23 @@ QueryPlanner::TripleGraph QueryPlanner::createTripleGraph(
            absl::StrSplit(sv.substr(1, sv.size() - 2), ' ')) {
         std::string s{ad_utility::utf8ToLower(term)};
         potentialTermsForCvar[t.s_.getVariable()].push_back(s);
+        if (activeGraphVariable_.has_value() ||
+            activeDatasetClauses_.defaultGraphs_.has_value()) {
+          AD_THROW(
+              "contains-word is not allowed inside GRAPH clauses or in queries "
+              "with FROM/FROM NAMED clauses.");
+        }
         addNodeToTripleGraph(
-            TripleGraph::Node(tg._nodeStorage.size(), t.s_.getVariable(), s, t),
+            TripleGraph::Node{tg._nodeStorage.size(), t.s_.getVariable(), s, t},
             tg);
         numNodesInTripleGraph++;
       }
     } else if (t.p_.iri_ == CONTAINS_ENTITY_PREDICATE) {
       entityTriples.push_back(&t);
     } else {
-      addNodeToTripleGraph(TripleGraph::Node(tg._nodeStorage.size(), t), tg);
+      addNodeToTripleGraph(
+          TripleGraph::Node{tg._nodeStorage.size(), t, activeGraphVariable_},
+          tg);
       numNodesInTripleGraph++;
     }
   }

diff --git a/src/engine/QueryPlanner.h b/src/engine/QueryPlanner.h
@@ -51,7 +51,9 @@ class QueryPlanner {
     TripleGraph(const TripleGraph& other, vector<size_t> keepNodes);
 
     struct Node {
-      Node(size_t id, SparqlTriple t) : id_(id), triple_(std::move(t)) {
+      Node(size_t id, SparqlTriple t,
+           std::optional<Variable> graphVariable = std::nullopt)
+          : id_(id), triple_(std::move(t)) {
         if (isVariable(triple_.s_)) {
           _variables.insert(triple_.s_.getVariable());
         }
@@ -61,6 +63,9 @@ class QueryPlanner {
         if (isVariable(triple_.o_)) {
           _variables.insert(triple_.o_.getVariable());
         }
+        if (graphVariable.has_value()) {
+          _variables.insert(std::move(graphVariable).value());
+        }
       }
 
       Node(size_t id, Variable cvar, std::string word, SparqlTriple t)

diff --git a/test/QueryPlannerTest.cpp b/test/QueryPlannerTest.cpp
@@ -2981,3 +2981,60 @@ TEST(QueryPlanner, Exists) {
       "GRAPH ?g { ?u ?v ?c}}}",
       filter);
 }
+
+// _____________________________________________________________________________
+TEST(QueryPlanner, ContainsWordInGraphClause) {
+  {
+    auto qp = makeQueryPlanner();
+    auto query = SparqlParser::parseQuery(
+        "SELECT * { GRAPH ?g { ?s "
+        "<http://qlever.cs.uni-freiburg.de/builtin-functions/contains-word> "
+        "\"Test\" } }");
+    AD_EXPECT_THROW_WITH_MESSAGE_AND_TYPE(
+        qp.createExecutionTree(query),
+        ::testing::HasSubstr(
+            "contains-word is not allowed inside GRAPH clauses "
+            "or in queries with FROM/FROM NAMED clauses."),
+        ad_utility::Exception);
+  }
+  {
+    auto qp = makeQueryPlanner();
+    auto query = SparqlParser::parseQuery(
+        "SELECT * { GRAPH <my-iri> { ?s "
+        "<http://qlever.cs.uni-freiburg.de/builtin-functions/contains-word> "
+        "\"Test\" } }");
+    AD_EXPECT_THROW_WITH_MESSAGE_AND_TYPE(
+        qp.createExecutionTree(query),
+        ::testing::HasSubstr(
+            "contains-word is not allowed inside GRAPH clauses "
+            "or in queries with FROM/FROM NAMED clauses."),
+        ad_utility::Exception);
+  }
+  {
+    auto qp = makeQueryPlanner();
+    auto query = SparqlParser::parseQuery(
+        "SELECT * FROM <my-iri> WHERE { ?s "
+        "<http://qlever.cs.uni-freiburg.de/builtin-functions/contains-word> "
+        "\"Test\" }");
+    AD_EXPECT_THROW_WITH_MESSAGE_AND_TYPE(
+        qp.createExecutionTree(query),
+        ::testing::HasSubstr(
+            "contains-word is not allowed inside GRAPH clauses "
+            "or in queries with FROM/FROM NAMED clauses."),
+        ad_utility::Exception);
+  }
+}
+
+// _____________________________________________________________________________
+TEST(QueryPlanner, UnconnectedComponentsInGraphClause) {
+  h::expect("SELECT * WHERE { GRAPH ?g { ?s1 ?p1 ?o1 . ?s2 ?p2 ?o2 } }",
+            h::Join(h::Sort(h::IndexScanFromStrings("?s1", "?p1", "?o1", {}, {},
+                                                    {Variable{"?g"}}, {3})),
+                    h::Sort(h::IndexScanFromStrings("?s2", "?p2", "?o2", {}, {},
+                                                    {Variable{"?g"}}, {3}))));
+  // Sanity check case without a GRAPH clause
+  h::expect(
+      "SELECT * WHERE { ?s1 ?p1 ?o1 . ?s2 ?p2 ?o2 }",
+      h::CartesianProductJoin(h::IndexScanFromStrings("?s1", "?p1", "?o1"),
+                              h::IndexScanFromStrings("?s2", "?p2", "?o2")));
+}