Fix issue 8757: Throw syntax error on invalid code (#1378)

* Fix issue 8757: Throw syntax error on invalid code

* Fix FP with lambda returns

* Remove double percent

* Check more keywords

* Skip preprocessor directives

* Check for valid PP directive

* Fix preprocessor check

* Dont check for preprocessor directives

lib/tokenize.cpp

@@ -8493,6 +8493,12 @@ void Tokenizer::findGarbageCode() const
             if (Token::Match(tok, "> %cop%"))
                 continue;
         }
+        if (Token::Match(tok, "%or%|%oror%|==|!=|+|-|/|!|>=|<=|~|++|--|::|sizeof|throw|decltype|typeof {|if|else|try|catch|while|do|for|return|switch|break|namespace"))
+            syntaxError(tok);
+        if (Token::Match(tok, "( %any% )") && tok->next()->isKeyword() && !Token::simpleMatch(tok->next(), "void"))
+            syntaxError(tok);
+        if (Token::Match(tok, "%num%|%bool%|%char%|%str% %num%|%bool%|%char%|%str%") && !Token::Match(tok, "%str% %str%"))
+            syntaxError(tok);
         if (Token::Match(tok, "%assign% typename|class %assign%"))
             syntaxError(tok);
         if (Token::Match(tok, "%cop%|=|,|[ %or%|%oror%|/|%"))

test/testgarbage.cpp

@@ -230,6 +230,7 @@ private:
         TEST_CASE(garbageCode197); // #8385
         TEST_CASE(garbageCode198); // #8383
         TEST_CASE(garbageCode199); // #8752
+        TEST_CASE(garbageCode200); // #8757
 
         TEST_CASE(garbageCodeFuzzerClientMode1); // test cases created with the fuzzer client, mode 1
 
@@ -975,7 +976,7 @@ private:
     }
 
     void garbageCode131() {
-        checkCode("( void ) { ( ) } ( ) / { ( ) }");
+        ASSERT_THROW(checkCode("( void ) { ( ) } ( ) / { ( ) }"), InternalError);
         // actually the invalid code should trigger an syntax error...
     }
 
@@ -1545,6 +1546,11 @@ private:
         checkCode("d f(){e n00e0[]n00e0&""0+f=0}");
     }
 
+    // #8757
+    void garbageCode200() {
+        ASSERT_THROW(checkCode("(){e break,{(case)!{e:[]}}}"), InternalError);
+    }
+
     void syntaxErrorFirstToken() {
         ASSERT_THROW(checkCode("&operator(){[]};"), InternalError); // #7818
         ASSERT_THROW(checkCode("*(*const<> (size_t); foo) { } *(*const (size_t)() ; foo) { }"), InternalError); // #6858
@@ -1599,6 +1605,10 @@ private:
         // case must be inside switch block
         ASSERT_THROW(checkCode("void f() { switch (a) {}; case 1: }"), InternalError); // #8184
         ASSERT_THROW(checkCode("struct V : { public case {} ; struct U : U  void { V *f (int x) (x) } }"), InternalError); // #5120
+        ASSERT_THROW(checkCode("void f() { 0 0; }"), InternalError);
+        ASSERT_THROW(checkCode("void f() { true 0; }"), InternalError);
+        ASSERT_THROW(checkCode("void f() { 'a' 0; }"), InternalError);
+        ASSERT_THROW(checkCode("void f() { 1 \"\"; }"), InternalError);
     }
 
     void enumTrailingComma() {

test/testtokenize.cpp

@@ -8641,6 +8641,8 @@ private:
         ASSERT_NO_THROW(tokenizeAndStringify("void f() { do switch (a) {} while (1); }"))
         ASSERT_NO_THROW(tokenizeAndStringify("void f() { label: switch (a) {} }"));
         ASSERT_NO_THROW(tokenizeAndStringify("void f() { UNKNOWN_MACRO if (a) {} }"))
+        ASSERT_NO_THROW(tokenizeAndStringify("void f() { []() -> int * {}; }"));
+        ASSERT_NO_THROW(tokenizeAndStringify("void f() { const char* var = \"1\" \"2\"; }"));
         // TODO ASSERT_NO_THROW(tokenizeAndStringify("void f() { MACRO(switch); }"));
         // TODO ASSERT_NO_THROW(tokenizeAndStringify("void f() { MACRO(x,switch); }"));