Fix rescanning when a macro expansion result forms a new function-like macro call (#682)

Author: pfultz2

simplecpp.cpp

@@ -1707,7 +1707,7 @@ namespace simplecpp {
                 if (output2.cfront() != output2.cback() && macro2tok->str() == this->name())
                     break;
                 const MacroMap::const_iterator macro = macros.find(macro2tok->str());
-                if (macro == macros.end() || !macro->second.functionLike())
+                if (macro == macros.end() || !macro->second.functionLike() || macro2tok->isExpandedFrom(&macro->second))
                     break;
                 TokenList rawtokens2(inputFiles);
                 const Location loc(macro2tok->location);
@@ -2161,39 +2161,37 @@ namespace simplecpp {
             return functionLike() ? parametertokens2.back()->next : nameTokInst->next;
         }
 
-        const Token *recursiveExpandToken(TokenList &output, TokenList &temp, const Location &loc, const Token *tok, const MacroMap &macros, const std::set<TokenString> &expandedmacros, const std::vector<const Token*> &parametertokens) const {
-            if (!temp.cback() || !temp.cback()->name || !tok->next || tok->next->op != '(') {
-                output.takeTokens(temp);
-                return tok->next;
-            }
-
-            if (!sameline(tok, tok->next)) {
-                output.takeTokens(temp);
-                return tok->next;
-            }
-
+        /** Returns the macro to expand when the last token of @p temp is the name of a
+         *  function-like macro and the tokens after @p tok supply its arguments; nullptr otherwise */
+        static const Macro *rescanMacro(const TokenList &temp, const Token *tok, const MacroMap &macros, const std::set<TokenString> &expandedmacros) {
+            if (!temp.cback() || !temp.cback()->name || !sameline(tok, tok->next) || tok->next->op != '(')
+                return nullptr;
             const MacroMap::const_iterator it = macros.find(temp.cback()->str());
-            if (it == macros.end() || expandedmacros.find(temp.cback()->str()) != expandedmacros.end()) {
-                output.takeTokens(temp);
-                return tok->next;
-            }
+            if (it == macros.end() || expandedmacros.find(temp.cback()->str()) != expandedmacros.end())
+                return nullptr;
+            if (!it->second.functionLike() || temp.cback()->isExpandedFrom(&it->second))
+                return nullptr;
+            return &it->second;
+        }
 
-            const Macro &calledMacro = it->second;
-            if (!calledMacro.functionLike()) {
+        const Token *recursiveExpandToken(TokenList &output, TokenList &temp, const Location &loc, const Token *tok, const MacroMap &macros, const std::set<TokenString> &expandedmacros, const std::vector<const Token*> &parametertokens) const {
+            // Expand while the expansion result ends with the name of a function-like
+            // macro whose arguments are supplied by the tokens that follow it. Each round
+            // consumes that macro call from the token stream, so tok always advances.
+            while (const Macro * const calledMacro = rescanMacro(temp, tok, macros, expandedmacros)) {
+                TokenList temp2(files);
+                temp2.push_back(new Token(temp.cback()->str(), tok->location));
+
+                const Token * const tok2 = appendTokens(temp2, loc, tok->next, macros, expandedmacros, parametertokens);
+                if (!tok2)
+                    break;
                 output.takeTokens(temp);
-                return tok->next;
+                output.deleteToken(output.back());
+                calledMacro->expand(temp, loc, temp2.cfront(), macros, expandedmacros);
+                tok = tok2;
             }
-
-            TokenList temp2(files);
-            temp2.push_back(new Token(temp.cback()->str(), tok->location));
-
-            const Token * const tok2 = appendTokens(temp2, loc, tok->next, macros, expandedmacros, parametertokens);
-            if (!tok2)
-                return tok->next;
             output.takeTokens(temp);
-            output.deleteToken(output.back());
-            calledMacro.expand(output, loc, temp2.cfront(), macros, expandedmacros);
-            return tok2->next;
+            return tok->next;
         }
 
         const Token *expandToken(TokenList &output, const Location &loc, const Token *tok, const MacroMap &macros, const std::set<TokenString> &expandedmacros, const std::vector<const Token*> &parametertokens) const {

test.cpp

@@ -949,6 +949,46 @@ static void define23() // #40
                   "unsigned A , B ;", preprocess(code));
 }
 
+static void define24()
+{
+    // an expansion result that is a function-like macro name must be rescanned
+    // repeatedly against the tokens that follow it
+    const char code[] = "#define a(b, c) c\n"
+                        "#define d() a\n"
+                        "#define g(e) h(e, ) h(e, )\n"
+                        "#define h(e, b) d()(, e)()\n"
+                        "#define i()\n"
+                        "g(i)\n";
+    ASSERT_EQUALS("", preprocess(code));
+}
+
+static void define25()
+{
+    // a macro name that came from expanding that same macro must not be
+    // re-expanded when rescanned with the tokens that follow it
+    const char code[] = "#define f() f\n"
+                        "#define wrap(x) x()\n"
+                        "wrap(f())\n";
+    ASSERT_EQUALS("\n"
+                  "\n"
+                  "f ( )", preprocess(code));
+}
+
+static void define26()
+{
+    // a macro name that came from expanding that same macro must not be
+    // re-expanded with arguments taken from the raw token stream
+    const char code[] = "#define f() f\n"
+                        "f()()\n";
+    ASSERT_EQUALS("\n"
+                  "f ( )", preprocess(code));
+
+    const char code2[] = "#define f() f\n"
+                         "f()()()\n";
+    ASSERT_EQUALS("\n"
+                  "f ( ) ( )", preprocess(code2));
+}
+
 
 static void define_invalid_1()
 {
@@ -4520,6 +4560,9 @@ static void runTests(int argc, char **argv, Input input)
     TEST_CASE(define21); // #66
     TEST_CASE(define22); // #40
     TEST_CASE(define23); // #40
+    TEST_CASE(define24);
+    TEST_CASE(define25);
+    TEST_CASE(define26);
     TEST_CASE(define_invalid_1);
     TEST_CASE(define_invalid_2);
     TEST_CASE(define_invalid_3);