Fix: bnode and variable validation (#306)

Author: liss-h

src/rdf4cpp/BlankNode.cpp

@@ -91,7 +91,7 @@ std::ostream &operator<<(std::ostream &os, BlankNode const &bnode) {
 
 void BlankNode::validate(std::string_view v) {
     using namespace util::char_matcher_detail;
-    static constexpr auto first_matcher = ASCIINumMatcher{} | PNCharsBaseMatcher;
+    static constexpr auto first_matcher = ASCIINumMatcher{} | PNCharsUMatcher;
     auto r = v | una::views::utf8;
     auto it = r.begin();
     if (it == r.end()) {
@@ -102,7 +102,7 @@ void BlankNode::validate(std::string_view v) {
     }
     auto lastchar = *it;
     ++it;
-    static constexpr auto pn_matcher = PNCharsMatcher | ASCIIPatternMatcher(".-");
+    static constexpr auto pn_matcher = PNCharsMatcher | ASCIIPatternMatcher{"."};
     while (it != r.end()) {
         if (!pn_matcher.match(*it))
         {

src/rdf4cpp/query/Variable.cpp

@@ -10,7 +10,7 @@ Variable::Variable() noexcept : Node{storage::identifier::NodeBackendHandle{{},
 }
 
 Variable::Variable(std::string_view name, bool anonymous, storage::DynNodeStoragePtr node_storage)
-    : Variable{make_unchecked((validate(name), name), anonymous, node_storage)} {}
+    : Variable{make_unchecked((validate(name, anonymous), name), anonymous, node_storage)} {}
 
 Variable::Variable(storage::identifier::NodeBackendHandle handle) noexcept : Node{handle} {}
 
@@ -108,9 +108,13 @@ std::ostream &operator<<(std::ostream &os, Variable const &variable) {
     return os;
 }
 
-void Variable::validate(std::string_view n) {
+void Variable::validate(std::string_view n, bool anonymous) {
+    if (anonymous) {
+        return BlankNode::validate(n);
+    }
+
     using namespace util::char_matcher_detail;
-    static constexpr auto first_matcher = ASCIINumMatcher{} | PNCharsBaseMatcher;
+    static constexpr auto first_matcher = ASCIINumMatcher{} | PNCharsUMatcher;
     auto r = n | una::views::utf8;
     auto it = r.begin();
     if (it == r.end()) {
@@ -120,9 +124,10 @@ void Variable::validate(std::string_view n) {
         throw InvalidNode(std::format("invalid blank node label {}", n));
     }
     ++it;
+
+    static constexpr auto matcher = ASCIINumMatcher{} | PNCharsUMatcher | PNChars_UnicodePartMatcher{};
     while (it != r.end()) {
-        if (!PNCharsMatcher.match(*it))
-        {
+        if (!matcher.match(*it)) {
             throw InvalidNode(std::format("invalid blank node label {}", n));
         }
         ++it;

src/rdf4cpp/query/Variable.hpp

@@ -46,9 +46,10 @@ struct Variable : Node {
     /**
      * Validates that the given name is a valid Variable name
      * @param var_name name to check
+     * @param anonymous if the variable is supposed to be anonymous
      * @throws ParsingError if the variable name is not valid
      */
-    static void validate(std::string_view var_name);
+    static void validate(std::string_view var_name, bool anonymous = false);
 
     [[nodiscard]] bool is_anonymous() const;
 

src/rdf4cpp/util/CharMatcher.hpp

@@ -264,12 +264,11 @@ struct IPrivateMatcher {
 };
 
 /**
- * matches PN_CHARS_BASE of the Turtle/SPARQL specification, without ASCII alpha (or with ASCIIAlphaMatcher).
+ * Matches the unicode part (the characters listed as numbers) of PN_CHARS_BASE of the Turtle/SPARQL specification
  */
-struct PNCharsBase_UniMatcher {
+struct PNCharsBase_UnicodePartMatcher {
     [[nodiscard]] static constexpr bool match(int c) noexcept {
-        return  c == '_' ||
-               (c >= 0xC0 && c <= 0xD6) ||
+        return (c >= 0xC0 && c <= 0xD6) ||
                (c >= 0xD8 && c <= 0xF6) ||
                (c >= 0xF8 && c <= 0x02FF) ||
                (c >= 0x0370 && c <= 0x037D) ||
@@ -296,12 +295,17 @@ struct PNCharsBase_UniMatcher {
 /**
  * matches PN_CHARS_BASE of the Turtle/SPARQL specification
  */
-constexpr auto PNCharsBaseMatcher = ASCIIAlphaMatcher{} | PNCharsBase_UniMatcher{};
+constexpr auto PNCharsBaseMatcher = ASCIIAlphaMatcher{} | PNCharsBase_UnicodePartMatcher{};
 
 /**
- * matches PN_CHARS of the Turtle/SPARQL specification, without ASCII num and PN_CHARS_BASE.
+ * matches PN_CHARS_U of the Turtle/SPARQL specificiation
  */
-struct PNChars_UniMatcher {
+constexpr auto PNCharsUMatcher = ASCIIPatternMatcher{"_"} | PNCharsBaseMatcher;
+
+/**
+ * Matches the unicode part (the characters listed as numbers) of PN_CHARS of the Turtle/SPARQL specification
+ */
+struct PNChars_UnicodePartMatcher {
     [[nodiscard]] static constexpr bool match(int c) noexcept {
         return c == 0xB7 ||
                (c >= 0x0300 && c <= 0x036F) ||
@@ -321,7 +325,7 @@ struct PNChars_UniMatcher {
 /**
  * matches PN_CHARS of the Turtle/SPARQL specification.
  */
-constexpr auto PNCharsMatcher = ASCIINumMatcher{} | PNCharsBaseMatcher | PNChars_UniMatcher{};
+constexpr auto PNCharsMatcher = ASCIINumMatcher{} | ASCIIPatternMatcher{"-"} | PNCharsUMatcher | PNChars_UnicodePartMatcher{};
 
 /**
   * iterates over s and tries to match all in m.

tests/query/tests_Variable.cpp

@@ -61,13 +61,13 @@ TEST_CASE("Variable::find") {
 }
 
 TEST_CASE("Variable validity") {
-    query::Variable n{};
-    CHECK_THROWS_AS(n = query::Variable::make_named("\U00000312_not_first"), InvalidNode);
-    CHECK_THROWS_AS(n = query::Variable::make_anonymous("no-"), InvalidNode);
-    CHECK_THROWS_AS(n = query::Variable::make_anonymous("-no"), InvalidNode);
-    CHECK_THROWS_AS(n = query::Variable("may_not_contain."), InvalidNode);
+    CHECK_THROWS_AS((void) query::Variable::make_named("\U00000312_not_first"), InvalidNode);
+    CHECK_THROWS_AS((void) query::Variable::make_named("no-"), InvalidNode);
+    CHECK(query::Variable::make_anonymous("no-").name() == "no-");
+    CHECK_THROWS_AS((void) query::Variable::make_named("-no"), InvalidNode);
+    CHECK_THROWS_AS((void) query::Variable::make_anonymous("-no"), InvalidNode);
+    CHECK_THROWS_AS((void) query::Variable("may_not_contain."), InvalidNode);
     CHECK(query::Variable::make_unchecked("may_not_contain.").name() == "may_not_contain.");
     CHECK(query::Variable::make_named("012_numbers_too567").name() == "012_numbers_too567");
     CHECK(query::Variable::make_named("\U0001f34cthrow_some_unicode_at_it\U0001f34c").name() == "\U0001f34cthrow_some_unicode_at_it\U0001f34c");
-    CHECK(n == query::Variable{});
 }