Tokenizer: Write syntax error if there is C++ code in C file.

lib/tokenize.cpp

@@ -3318,6 +3318,9 @@ bool Tokenizer::simplifyTokenList1(const char FileName[])
         }
     }
 
+    // Is there C++ code in C file?
+    validateC();
+
     // remove MACRO in variable declaration: MACRO int x;
     removeMacroInVarDecl();
 
@@ -7617,6 +7620,12 @@ void Tokenizer::syntaxError(const Token *tok, char c) const
                             InternalError::SYNTAX);
 }
 
+void Tokenizer::syntaxErrorC(const Token *tok, const std::string &what) const
+{
+    printDebugOutput(0);
+    throw InternalError(tok, "Code '"+what+"' is invalid C code. Use --std or --language to configure the language.", InternalError::SYNTAX);
+}
+
 void Tokenizer::unhandled_macro_class_x_y(const Token *tok) const
 {
     reportError(tok,
@@ -7996,6 +8005,24 @@ void Tokenizer::simplifyComma()
 }
 
 
+void Tokenizer::validateC() const
+{
+    if (!isC())
+        return;
+    for (const Token *tok = tokens(); tok; tok = tok->next()) {
+        if (tok->previous() && !Token::Match(tok->previous(), "[;{}]"))
+            continue;
+        if (Token::simpleMatch(tok, "using namespace std ;"))
+            syntaxErrorC(tok, "using namespace std");
+        if (Token::Match(tok, "template < class %name% [,>]"))
+            syntaxErrorC(tok, "template<...");
+        if (Token::Match(tok, "%name% :: %name%"))
+            syntaxErrorC(tok, tok->str() + tok->strAt(1) + tok->strAt(2));
+        if (Token::Match(tok, "class|namespace %name% [:{]"))
+            syntaxErrorC(tok, tok->str() + tok->strAt(1) + tok->strAt(2));
+    }
+}
+
 void Tokenizer::validate() const
 {
     std::stack<const Token *> linkTokens;

lib/tokenize.h

@@ -568,11 +568,19 @@ public:
     /** Syntax error. Example: invalid number of ')' */
     void syntaxError(const Token *tok, char c) const;
 
+    /** Syntax error. C++ code in C file. */
+    void syntaxErrorC(const Token *tok, const std::string &what) const;
+
 private:
 
     /** Report that there is an unhandled "class x y {" code */
     void unhandled_macro_class_x_y(const Token *tok) const;
 
+    /**
+     * Is there C++ code in C file?
+     */
+    void validateC() const;
+
     /**
      * assert that tokens are ok - used during debugging for example
      * to catch problems in simplifyTokenList1/2.

test/testgarbage.cpp

@@ -130,7 +130,6 @@ private:
         TEST_CASE(garbageCode90);
         TEST_CASE(garbageCode91);
         TEST_CASE(garbageCode92);
-        TEST_CASE(garbageCode93);
         TEST_CASE(garbageCode94);
         TEST_CASE(garbageCode95);
         TEST_CASE(garbageCode96);
@@ -763,10 +762,6 @@ private:
         ASSERT_THROW(checkCode("template < typename _Tp ( ( ) ; _Tp ) , decltype > { } { ( ) ( ) }"), InternalError); // do not crash
     }
 
-    void garbageCode93() { // #6800
-        checkCode(" namespace A { } class A{ { }} class A : T ;", false); // do not crash
-    }
-
     void garbageCode94() { // #6803
         //checkCode("typedef long __m256i __attribute__ ( ( ( ) ) )[ ; ( ) { } typedef __m256i __attribute__ ( ( ( ) ) ) < ] ( ) { ; }");
         ASSERT_THROW(checkCode("typedef long __m256i __attribute__ ( ( ( ) ) )[ ; ( ) { } typedef __m256i __attribute__ ( ( ( ) ) ) < ] ( ) { ; }"), InternalError);
@@ -1326,7 +1321,7 @@ private:
 
     void garbageCode164() {
         //7234
-        checkCode("class d{k p;}(){d::d():B<()}", false);
+        checkCode("class d{k p;}(){d::d():B<()}");
     }
 
     void garbageCode165() {
@@ -1336,7 +1331,7 @@ private:
 
     void garbageCode167() {
         //7237
-        checkCode("class D00i000{:D00i000::}i", false);
+        ASSERT_THROW(checkCode("class D00i000{:D00i000::}i"),InternalError);
     }
 
     void garbageCode168() {

test/testleakautovar.cpp

@@ -864,7 +864,7 @@ private:
               "    ::exit(0);\n"
               "  }"
               "  free(p);\n"
-              "}");
+              "}", true);
         ASSERT_EQUALS("", errout.str());
 
         check("void f() {\n"
@@ -874,7 +874,7 @@ private:
               "    std::exit(0);\n"
               "  }"
               "  free(p);\n"
-              "}");
+              "}", true);
         ASSERT_EQUALS("", errout.str());
     }
 

test/testother.cpp

@@ -1104,15 +1104,8 @@ private:
         std::istringstream istr(code);
         tokenizerCpp.tokenize(istr, "test.cpp");
 
-        Tokenizer tokenizerC(&settings, this);
-        std::istringstream istr2(code);
-        tokenizerC.tokenize(istr2, "test.c");
-
         CheckOther checkOtherCpp(&tokenizerCpp, &settings, this);
         checkOtherCpp.warningOldStylePointerCast();
-
-        CheckOther checkOtherC(&tokenizerC, &settings, this);
-        checkOtherC.warningOldStylePointerCast();
     }
 
     void oldStylePointerCast() {

test/testtokenize.cpp

@@ -64,6 +64,8 @@ private:
         TEST_CASE(tokenize32);  // #5884 (fsanitize=undefined: left shift of negative value -10000 in lib/templatesimplifier.cpp:852:46)
         TEST_CASE(tokenize33);  // #5780 Various crashes on valid template code
 
+        TEST_CASE(validate);
+
         TEST_CASE(syntax_case_default);
 
         TEST_CASE(foreach);     // #3690
@@ -780,6 +782,13 @@ private:
         tokenizeAndStringify(code, true);
     }
 
+    void validate() {
+        // C++ code in C file
+        ASSERT_THROW(tokenizeAndStringify(";using namespace std;",false,false,Settings::Native,"test.c"), InternalError);
+        ASSERT_THROW(tokenizeAndStringify(";std::map<int,int> m;",false,false,Settings::Native,"test.c"), InternalError);
+        ASSERT_THROW(tokenizeAndStringify(";template<class T> class X { };",false,false,Settings::Native,"test.c"), InternalError);
+    }
+
     void syntax_case_default() { // correct syntax
         tokenizeAndStringify("void f() {switch (n) { case 0: z(); break;}}");
         ASSERT_EQUALS("", errout.str());