Support BASE declarations in SPARQL queries

src/parser/sparqlParser/SparqlQleverVisitor.cpp

@@ -1163,8 +1163,20 @@ TripleComponent::Iri Visitor::visit(Parser::IriContext* ctx) {
 }
 
 // ____________________________________________________________________________________
-string Visitor::visit(Parser::IrirefContext* ctx) {
-  return RdfEscaping::unescapeIriref(ctx->getText());
+string Visitor::visit(Parser::IrirefContext* ctx) const {
+  auto string = ctx->getText();
+  auto removeIriBrackets = [](std::string_view string) {
+    AD_CORRECTNESS_CHECK(string.starts_with('<'));
+    AD_CORRECTNESS_CHECK(string.ends_with('>'));
+    return string.substr(1, string.size() - 2);
+  };
+  // Handle relative IRIs, technically relative IRIs are not allowed when no
+  // BASE declaration is present, but we use them a lot for unit tests.
+  if (!baseIri_.empty() && !isAbsoluteIri(string)) {
+    return absl::StrCat("<", removeIriBrackets(baseIri_),
+                        removeIriBrackets(string), ">");
+  }
+  return string;
 }
 
 // ____________________________________________________________________________________
@@ -1212,8 +1224,15 @@ DatasetClause SparqlQleverVisitor::visit(Parser::UsingClauseContext* ctx) {
 
 // ____________________________________________________________________________________
 void Visitor::visit(Parser::PrologueContext* ctx) {
-  visitVector(ctx->baseDecl());
-  visitVector(ctx->prefixDecl());
+  for (auto* child : ctx->children) {
+    if (auto* baseDecl = dynamic_cast<Parser::BaseDeclContext*>(child)) {
+      visit(baseDecl);
+    } else {
+      auto* prefixDecl = dynamic_cast<Parser::PrefixDeclContext*>(child);
+      AD_CORRECTNESS_CHECK(prefixDecl);
+      visit(prefixDecl);
+    }
+  }
   // Remember the whole prologue (we need this when we encounter a SERVICE
   // clause, see `visit(ServiceGraphPatternContext*)` below.
   if (ctx->getStart() && ctx->getStop()) {
@@ -1222,8 +1241,17 @@ void Visitor::visit(Parser::PrologueContext* ctx) {
 }
 
 // ____________________________________________________________________________________
-void Visitor::visit(const Parser::BaseDeclContext* ctx) {
-  reportNotSupported(ctx, "BASE declarations are");
+bool Visitor::isAbsoluteIri(std::string_view string) {
+  return ctre::starts_with<"<[A-Za-z]*[A-Za-z0-9+-.]:">(string);
+}
+
+// ____________________________________________________________________________________
+void Visitor::visit(Parser::BaseDeclContext* ctx) {
+  auto rawIri = ctx->iriref()->getText();
+  if (!isAbsoluteIri(ctx->iriref()->getText())) {
+    reportError(ctx, "The base IRI must be an absolute IRI, was: " + rawIri);
+  }
+  baseIri_ = visit(ctx->iriref());
 }
 
 // ____________________________________________________________________________________

src/parser/sparqlParser/SparqlQleverVisitor.h

@@ -80,6 +80,7 @@ class SparqlQleverVisitor {
   // `addVisibleVariable`.
   std::vector<Variable> visibleVariables_{};
   PrefixMap prefixMap_{};
+  std::string baseIri_ = "";
   // We need to remember the prologue (prefix declarations) when we encounter it
   // because we need it when we encounter a SERVICE query. When there is no
   // prologue, this string simply remains empty.
@@ -121,8 +122,11 @@ class SparqlQleverVisitor {
   // ___________________________________________________________________________
   void visit(Parser::PrologueContext* ctx);
 
+  // Returns true if the given string is an absolute IRI.
+  static bool isAbsoluteIri(std::string_view);
+
   // ___________________________________________________________________________
-  [[noreturn]] static void visit(const Parser::BaseDeclContext* ctx);
+  void visit(Parser::BaseDeclContext* ctx);
 
   // ___________________________________________________________________________
   void visit(Parser::PrefixDeclContext* ctx);
@@ -469,7 +473,7 @@ class SparqlQleverVisitor {
 
   TripleComponent::Iri visit(Parser::IriContext* ctx);
 
-  static string visit(Parser::IrirefContext* ctx);
+  string visit(Parser::IrirefContext* ctx) const;
 
   string visit(Parser::PrefixedNameContext* ctx);
 

test/SparqlParserTest.cpp

@@ -1349,3 +1349,37 @@ TEST(ParserTest, HandlesSurrogatesCorrectly) {
   // So writing unit tests for these cases is not possible without creating
   // semi-invalid UTF-8 strings.
 }
+
+// _____________________________________________________________________________
+TEST(ParserTest, BaseDeclaration) {
+  // Simple case
+  auto query1 = SparqlParser::parseQuery(
+      "BASE <http://example.org/> SELECT * WHERE { ?s <p> <test> }");
+  EXPECT_EQ(getFirstTriple(query1),
+            "{s: ?s, p: <http://example.org/p>, o: <http://example.org/test>}");
+  // Relative and absolute IRIs mixed
+  auto query2 = SparqlParser::parseQuery(
+      "BASE <http://example.org/> "
+      "SELECT * WHERE { ?s <p> <http://other.example.org/p> }");
+  EXPECT_EQ(
+      getFirstTriple(query2),
+      "{s: ?s, p: <http://example.org/p>, o: <http://other.example.org/p>}");
+
+  // Cascading prefixes
+  auto query3 = SparqlParser::parseQuery(
+      "BASE <http://example.org/> "
+      "PREFIX ex1: <ex1/> "
+      "BASE <http://other.example.org/> "
+      "PREFIX ex2: <ex2/> "
+      "BASE <http://alternative.example.org/> "
+      "SELECT * WHERE { ex2:hello <world> ex1:test }");
+  EXPECT_EQ(getFirstTriple(query3),
+            "{s: <http://other.example.org/ex2/hello>,"
+            " p: <http://alternative.example.org/world>,"
+            " o: <http://example.org/ex1/test>}");
+
+  AD_EXPECT_THROW_WITH_MESSAGE_AND_TYPE(
+      SparqlParser::parseQuery("BASE <http://example.com> BASE <relative> "
+                               "SELECT * WHERE { ?s ?p ?o }"),
+      ::testing::HasSubstr("absolute IRI"), InvalidSparqlQueryException);
+}