Fix a small bug in the Turtle parser (ad-freiburg#1394)

Fixes ad-freiburg#1392 . The problem occurred for typed literals with escaped characters.
Jonathan24680 · Jul 15, 2024 · 6a6a4b8 · 6a6a4b8
1 parent 67145fa
commit 6a6a4b8
Show file tree

Hide file tree

Showing 4 changed files with 22 additions and 12 deletions.
diff --git a/src/parser/RdfEscaping.cpp b/src/parser/RdfEscaping.cpp
@@ -41,12 +41,12 @@ std::string hexadecimalCharactersToUtf8(std::string_view hex) {
  * is not allowed is found.
  */
 template <bool acceptOnlyNumericEscapes, bool acceptOnlyBackslashAndNewline,
-          typename InputIterator, typename OutputIterator>
-void unescapeStringAndNumericEscapes(InputIterator beginIterator,
-                                     InputIterator endIterator,
+          typename OutputIterator>
+void unescapeStringAndNumericEscapes(std::string_view input,
                                      OutputIterator outputIterator) {
   static_assert(!(acceptOnlyNumericEscapes && acceptOnlyBackslashAndNewline));
-
+  auto beginIterator = input.begin();
+  auto endIterator = input.end();
   // Append the `character` to the output, but only if newlines/backslashes are
   // allowed via the configuration
   auto pushNewlineOrBackslash = [&outputIterator](
@@ -141,8 +141,9 @@ void unescapeStringAndNumericEscapes(InputIterator beginIterator,
       }
 
       default:
-        // should never happen
-        AD_FAIL();
+        AD_CONTRACT_CHECK(false,
+                          "Unsupported escape sequence found in RDF literal \"",
+                          input, "\"");
     }
     beginIterator = nextBackslashIterator + numCharactersFromInput;
   }
@@ -153,7 +154,7 @@ void unescapeStringAndNumericEscapes(InputIterator beginIterator,
 std::string unescapeNewlinesAndBackslashes(std::string_view literal) {
   std::string result;
   RdfEscaping::detail::unescapeStringAndNumericEscapes<false, true>(
-      literal.begin(), literal.end(), std::back_inserter(result));
+      literal, std::back_inserter(result));
   return result;
 }
 
@@ -165,7 +166,7 @@ std::string escapeNewlinesAndBackslashes(std::string_view literal) {
 // ____________________________________________________________________________
 static void literalUnescape(std::string_view input, std::string& res) {
   detail::unescapeStringAndNumericEscapes<false, false>(
-      input.begin(), input.end(), std::back_inserter(res));
+      input, std::back_inserter(res));
 }
 
 // ____________________________________________________________________________
@@ -223,7 +224,7 @@ static void unescapeIriWithoutBrackets(std::string_view input,
                                        std::string& res) {
   // Only numeric escapes are allowed for iriefs.
   RdfEscaping::detail::unescapeStringAndNumericEscapes<true, false>(
-      input.begin(), input.end(), std::back_inserter(res));
+      input, std::back_inserter(res));
 }
 
 // __________________________________________________________________________

diff --git a/src/parser/TurtleParser.cpp b/src/parser/TurtleParser.cpp
@@ -393,8 +393,8 @@ template <class T>
 TripleComponent TurtleParser<T>::literalAndDatatypeToTripleComponentImpl(
     std::string_view normalizedLiteralContent,
     const TripleComponent::Iri& typeIri, TurtleParser<T>* parser) {
-  auto literal =
-      TripleComponent::Literal::literalWithoutQuotes(normalizedLiteralContent);
+  auto literal = TripleComponent::Literal::literalWithNormalizedContent(
+      asNormalizedStringViewUnsafe(normalizedLiteralContent));
   std::string_view type = asStringViewUnsafe(typeIri.getContent());
 
   try {

diff --git a/test/RdfEscapingTest.cpp b/test/RdfEscapingTest.cpp
@@ -5,8 +5,8 @@
 
 #include <gtest/gtest.h>
 
+#include "./util/GTestHelpers.h"
 #include "parser/RdfEscaping.h"
-
 using namespace RdfEscaping;
 
 // ___________________________________________________________________________
@@ -43,6 +43,11 @@ TEST(RdfEscapingTest, normalizedContentFromLiteralOrIri) {
   ASSERT_EQ(f("\"bumm\"^^<http://www.mycustomiris.com/sometype>"), "bumm");
 }
 
+TEST(RdfEscapingTest, invalidEscapeThrows) {
+  AD_EXPECT_THROW_WITH_MESSAGE(
+      normalizeRDFLiteral("\"invalid\\Escape\""),
+      ::testing::HasSubstr("Unsupported escape sequence"));
+}
 // ___________________________________________________________________________
 TEST(RdfEscapingTest, escapeForXml) {
   ASSERT_EQ(escapeForXml("abc\n\t;"), "abc\n\t;");

diff --git a/test/TurtleParserTest.cpp b/test/TurtleParserTest.cpp
@@ -197,6 +197,10 @@ TEST(TurtleParserTest, rdfLiteral) {
   std::vector<TripleComponent> expected;
   literals.emplace_back(R"("simpleString")");
   expected.emplace_back(lit(R"("simpleString")"));
+  literals.emplace_back(R"("string\"with \\ escapes\n"^^<www.x.de>)");
+  expected.emplace_back(TripleComponent::Literal::fromEscapedRdfLiteral(
+      R"("string\"with \\ escapes\n")",
+      TripleComponent::Iri::fromIriref("<www.x.de>")));
   literals.emplace_back(R"("langtag"@en-gb)");
   expected.emplace_back(lit(R"("langtag")", "@en-gb"));
   literals.emplace_back("\"valueLong\"^^<www.someunknownType/integer>");