Merge pull request #396 from danmar/fix-395

danmar · Dec 11, 2024 · d4323f7 · d4323f7
2 parents bacd5bd + 0b6feab
commit d4323f7
Show file tree

Hide file tree

Showing 2 changed files with 39 additions and 4 deletions.
diff --git a/simplecpp.cpp b/simplecpp.cpp
@@ -110,6 +110,15 @@ template<class T> static std::string toString(T t)
     return ostr.str();
 }
 
+#ifdef SIMPLECPP_DEBUG_MACRO_EXPANSION
+static std::string locstring(const simplecpp::Location &loc)
+{
+    std::ostringstream ostr;
+    ostr << '[' << loc.file() << ':' << loc.line << ':' << loc.col << ']';
+    return ostr.str();
+}
+#endif
+
 static long long stringToLL(const std::string &s)
 {
     long long ret;
@@ -1528,6 +1537,10 @@ namespace simplecpp {
                              std::vector<std::string> &inputFiles) const {
             std::set<TokenString> expandedmacros;
 
+#ifdef SIMPLECPP_DEBUG_MACRO_EXPANSION
+            std::cout << "expand " << name() << " " << locstring(rawtok->location) << std::endl;
+#endif
+
             TokenList output2(inputFiles);
 
             if (functionLike() && rawtok->next && rawtok->next->op == '(') {
@@ -1797,6 +1810,10 @@ namespace simplecpp {
         const Token * expand(TokenList * const output, const Location &loc, const Token * const nameTokInst, const MacroMap &macros, std::set<TokenString> expandedmacros) const {
             expandedmacros.insert(nameTokInst->str());
 
+#ifdef SIMPLECPP_DEBUG_MACRO_EXPANSION
+            std::cout << "  expand " << name() << " " << locstring(defineLocation()) << std::endl;
+#endif
+
             usageList.push_back(loc);
 
             if (nameTokInst->str() == "__FILE__") {
@@ -2168,8 +2185,7 @@ namespace simplecpp {
 
             const bool canBeConcatenatedWithEqual = A->isOneOf("+-*/%&|^") || A->str() == "<<" || A->str() == ">>";
             const bool canBeConcatenatedStringOrChar = isStringLiteral_(A->str()) || isCharLiteral_(A->str());
-            if (!A->name && !A->number && A->op != ',' && !A->str().empty() && !canBeConcatenatedWithEqual && !canBeConcatenatedStringOrChar)
-                throw invalidHashHash::unexpectedToken(tok->location, name(), A);
+            const bool unexpectedA = (!A->name && !A->number && !A->str().empty() && !canBeConcatenatedWithEqual && !canBeConcatenatedStringOrChar);
 
             Token * const B = tok->next->next;
             if (!B->name && !B->number && B->op && !B->isOneOf("#="))
@@ -2187,6 +2203,9 @@ namespace simplecpp {
             const Token *nextTok = B->next;
 
             if (canBeConcatenatedStringOrChar) {
+                if (unexpectedA)
+                    throw invalidHashHash::unexpectedToken(tok->location, name(), A);
+
                 // It seems clearer to handle this case separately even though the code is similar-ish, but we don't want to merge here.
                 // TODO The question is whether the ## or varargs may still apply, and how to provoke?
                 if (expandArg(&tokensB, B, parametertokens)) {
@@ -2205,13 +2224,17 @@ namespace simplecpp {
                 if (expandArg(&tokensB, B, parametertokens)) {
                     if (tokensB.empty())
                         strAB = A->str();
-                    else if (varargs && A->op == ',') {
+                    else if (varargs && A->op == ',')
                         strAB = ",";
-                    } else {
+                    else if (varargs && unexpectedA)
+                        throw invalidHashHash::unexpectedToken(tok->location, name(), A);
+                    else {
                         strAB = A->str() + tokensB.cfront()->str();
                         tokensB.deleteToken(tokensB.front());
                     }
                 } else {
+                    if (unexpectedA)
+                        throw invalidHashHash::unexpectedToken(tok->location, name(), A);
                     strAB = A->str() + B->str();
                 }
 

diff --git a/test.cpp b/test.cpp
@@ -1421,6 +1421,17 @@ static void hashhash_null_stmt()
     ASSERT_EQUALS("\n\n\n\n1 ;", preprocess(code, &outputList));
 }
 
+static void hashhash_empty_va_args()
+{
+    // #395 hash hash with an empty __VA_ARGS__ in a macro
+    const char code[] =
+        "#define CAT(a, ...)        a##__VA_ARGS__\n"
+        "#define X(a, ...)          CAT(a)\n"
+        "#define LEVEL_2            (2)\n"
+        "X(LEVEL_2)\n";
+    ASSERT_EQUALS("\n\n\n( 2 )", preprocess(code));
+}
+
 static void hashhash_universal_character()
 {
     const char code[] =
@@ -3044,6 +3055,7 @@ int main(int argc, char **argv)
     TEST_CASE(hashhash_invalid_string_number);
     TEST_CASE(hashhash_invalid_missing_args);
     TEST_CASE(hashhash_null_stmt);
+    TEST_CASE(hashhash_empty_va_args);
     // C standard, 5.1.1.2, paragraph 4:
     //    If a character sequence that matches the syntax of a universal
     //    character name is produced by token concatenation (6.10.3.3),