Fix REPLACE function with replace patterns

src/engine/sparqlExpressions/SparqlExpressionValueGetters.cpp

@@ -90,6 +90,55 @@ std::optional<std::string> StringValueGetter::operator()(
   }
 }
 
+// ____________________________________________________________________________
+std::optional<std::string> ReplacementStringGetter::operator()(
+    Id id, const EvaluationContext* context) const {
+  std::optional<std::string> originalString =
+      StringValueGetter::operator()(id, context);
+  if (!originalString.has_value()) {
+    return originalString;
+  }
+  return convertToReplacementString(originalString.value());
+}
+
+// ____________________________________________________________________________
+std::optional<std::string> ReplacementStringGetter::operator()(
+    const LiteralOrIri& s, const EvaluationContext*) const {
+  return convertToReplacementString(asStringViewUnsafe(s.getContent()));
+}
+
+// ____________________________________________________________________________
+std::string ReplacementStringGetter::convertToReplacementString(
+    std::string_view view) {
+  std::string result;
+  // Rough estimate of the size of the result string.
+  result.reserve(view.size());
+  for (size_t i = 0; i < view.size(); i++) {
+    char c = view.at(i);
+    switch (c) {
+      case '$':
+        // Re2 used \1, \2, ... for backreferences, so we change $ to \.
+        result.push_back('\\');
+        break;
+      case '\\':
+        // "\$" is unescaped to "$"
+        if (i + 1 < view.size() && view.at(i + 1) == '$') {
+          result.push_back('$');
+          i++;
+        } else {
+          // Escape existing backslashes.
+          result.push_back(c);
+          result.push_back(c);
+        }
+        break;
+      default:
+        result.push_back(c);
+        break;
+    }
+  }
+  return result;
+}
+
 // ____________________________________________________________________________
 template <auto isSomethingFunction, auto prefix>
 Id IsSomethingValueGetter<isSomethingFunction, prefix>::operator()(

src/engine/sparqlExpressions/SparqlExpressionValueGetters.h

@@ -140,6 +140,20 @@ struct StringValueGetter : Mixin<StringValueGetter> {
     return std::string(asStringViewUnsafe(s.getContent()));
   }
 };
+// Similar to `StringValueGetter`, but correctly escapes and unescapes string
+// sequences.
+struct ReplacementStringGetter : StringValueGetter,
+                                 Mixin<ReplacementStringGetter> {
+  using Mixin<ReplacementStringGetter>::operator();
+  std::optional<std::string> operator()(ValueId,
+                                        const EvaluationContext*) const;
+
+  std::optional<std::string> operator()(const LiteralOrIri& s,
+                                        const EvaluationContext*) const;
+
+ private:
+  static std::string convertToReplacementString(std::string_view view);
+};
 
 // Boolean value getter that checks whether the given `Id` is a `ValueId` of the
 // given `datatype`.

src/engine/sparqlExpressions/StringExpressions.cpp

@@ -357,7 +357,7 @@ using StrBeforeExpression =
 
 using ReplaceExpression =
     StringExpressionImpl<3, decltype(replaceImpl), RegexValueGetter,
-                         StringValueGetter>;
+                         ReplacementStringGetter>;
 
 // CONCAT
 class ConcatExpression : public detail::VariadicExpression {

test/SparqlExpressionTest.cpp

@@ -1242,6 +1242,17 @@ TEST(SparqlExpression, ReplaceExpression) {
       idOrLitOrStringVec({"null", "Xs", "zwei", "drei", U, U}),
       std::tuple{idOrLitOrStringVec({"null", "eins", "zwei", "drei", U, U}),
                  IdOrLiteralOrIri{lit("e.[a-z]")}, IdOrLiteralOrIri{lit("X")}});
+  // A regex with replacement with substitutions
+  checkReplace(
+      idOrLitOrStringVec({"\"$1 \\\\2 A \\\\bc\"", "\"$1 \\\\2 DE \\\\f\""}),
+      std::tuple{idOrLitOrStringVec({"Abc", "DEf"}),
+                 IdOrLiteralOrIri{lit("([A-Z]+)")},
+                 IdOrLiteralOrIri{lit("\"\\\\$1 \\\\2 $1 \\\\\"")}});
+
+  checkReplace(idOrLitOrStringVec({"truebc", "truef"}),
+               std::tuple{idOrLitOrStringVec({"Abc", "DEf"}),
+                          IdOrLiteralOrIri{lit("([A-Z]+)")},
+                          IdOrLiteralOrIri{Id::makeFromBool(true)}});
 
   // Case-insensitive matching using the hack for google regex:
   checkReplace(idOrLitOrStringVec({"null", "xxns", "zwxx", "drxx"}),