Fixes #14921: Detect enums before simplifying platform types

Author: autoantwort

lib/tokenlist.cpp

@@ -2077,6 +2077,45 @@ bool TokenList::validateToken(const Token* tok) const
     return false;
 }
 
+/** Return whether tok is the "{" that starts an enumerator list */
+static bool isEnumStart(const Token* tok)
+{
+    if (!Token::simpleMatch(tok, "{"))
+        return false;
+    tok = tok->previous();
+    while (tok && (!tok->isKeyword() || Token::isStandardType(tok->str())) && Token::Match(tok, "%name%|::|:"))
+        tok = tok->previous();
+    if (Token::simpleMatch(tok, "class"))
+        tok = tok->previous();
+    return Token::simpleMatch(tok, "enum");
+}
+
+// Collect the tokens that are enumerator names being declared in some enum's member list,
+// e.g. the HANDLE in "enum E { HANDLE, ... };". Each enum body is scanned once, jumping over
+// nested brackets in constant-expressions via link(), so the total cost is bounded by the
+// combined size of all enum bodies -- not by unrelated code elsewhere in the file.
+static std::unordered_set<const Token*> findEnumeratorNames(const Token* front)
+{
+    std::unordered_set<const Token*> result;
+    for (const Token* tok = front; tok; tok = tok->next()) {
+        if (!isEnumStart(tok))
+            continue;
+        const Token* const bodyEnd = tok->link();
+        for (const Token* nameTok = tok->next(); nameTok && nameTok != bodyEnd;) {
+            if (nameTok->isName())
+                result.insert(nameTok);
+            while (nameTok != bodyEnd && nameTok->str() != ",") {
+                if (Token::Match(nameTok, "(|[|{"))
+                    nameTok = nameTok->link();
+                nameTok = nameTok->next();
+            }
+            if (nameTok != bodyEnd) // skip the ","
+                nameTok = nameTok->next();
+        }
+    }
+    return result;
+}
+
 void TokenList::simplifyPlatformTypes()
 {
     const bool isCPP11 = isCPP() && (mSettings.standards.cpp >= Standards::CPP11);
@@ -2143,6 +2182,7 @@ void TokenList::simplifyPlatformTypes()
     }
 
     const std::string platform_type(mSettings.platform.toString());
+    const std::unordered_set<const Token*> enumeratorNames = findEnumeratorNames(front());
 
     for (Token *tok = front(); tok; tok = tok->next()) {
         if (tok->tokType() != Token::eType && tok->tokType() != Token::eName)
@@ -2151,6 +2191,9 @@ void TokenList::simplifyPlatformTypes()
         const Library::PlatformType * const platformtype = mSettings.library.platform_type(tok->str(), platform_type);
 
         if (platformtype) {
+            if (enumeratorNames.count(tok) != 0) // don't replace an enumerator's name, e.g. HANDLE in "enum E { HANDLE, ... };"
+                continue;
+
             // check for namespace
             if (tok->strAt(-1) == "::") {
                 const Token * tok1 = tok->tokAt(-2);

test/testtokenize.cpp

@@ -400,6 +400,7 @@ class TestTokenizer : public TestFixture {
         TEST_CASE(multipleAssignment);
 
         TEST_CASE(platformWin);
+        TEST_CASE(platformWinEnumerator);
         TEST_CASE(platformWin32A);
         TEST_CASE(platformWin32W);
         TEST_CASE(platformWin32AStringCat); // ticket #5015
@@ -6264,6 +6265,31 @@ class TestTokenizer : public TestFixture {
         ASSERT_EQUALS(win32A, tokenizeAndStringify(code, settings_win32a));
     }
 
+    void platformWinEnumerator() {
+        // #11538 - HANDLE is a platform type (-> void*) but here it's an enumerator and must not be replaced
+        {
+            const char code[] = "enum class E { HANDLE, OTHER };";
+            ASSERT_EQUALS("enum class E { HANDLE , OTHER } ;", tokenizeAndStringify(code, settings_win32a));
+        }
+        // a platform type used within an enumerator's constant-expression is not a name and must still be replaced
+        {
+            const char code[] = "enum E { A = sizeof(HANDLE) };";
+            ASSERT_EQUALS("enum E { A = sizeof ( void * ) } ;", tokenizeAndStringify(code, settings_win32a));
+        }
+        // combines both: DWORD is the enum's underlying type (-> replaced), HANDLE is the enumerator name (-> kept),
+        // and the HANDLE inside sizeof() is part of the constant-expression (-> replaced)
+        {
+            const char code[] = "enum class T : DWORD { HANDLE = sizeof(HANDLE) };";
+            ASSERT_EQUALS("enum class T : unsigned long { HANDLE = sizeof ( void * ) } ;", tokenizeAndStringify(code, settings_win32a));
+        }
+        // a first naive fix walked backwards from every platform-type token to find an enclosing "{".
+        // this implementation failed on the following example, therefore test it.
+        {
+            const char code[] = "enum T { }; void test(int, HANDLE, int);";
+            ASSERT_EQUALS("enum T { } ; void test ( int , void * , int ) ;", tokenizeAndStringify(code, settings_win32a));
+        }
+    }
+
     void platformWin32A() {
         const char code[] = "wchar_t wc;"
                             "TCHAR c;"