Fix #3208 (Simplify pointer to standard type, C only)

The symbol database is unavailable during token simplification
and &data[0] might return something completely different for C++.

Moved code_is_c() from checkOther to Tokenizer.

lib/checkother.cpp

@@ -147,7 +147,7 @@ void CheckOther::clarifyCondition()
                     tok2 = tok2->link();
                 else if (Token::Match(tok2, "<|<=|==|!=|>|>=")) {
                     // This might be a template
-                    if (!code_is_c() && Token::Match(tok2->previous(), "%var% <"))
+                    if (!_tokenizer->code_is_c() && Token::Match(tok2->previous(), "%var% <"))
                         break;
 
                     clarifyConditionError(tok, tok->strAt(2) == "=", false);
@@ -1163,7 +1163,7 @@ void CheckOther::checkComparisonOfBoolWithInt()
             std::map<unsigned int, bool>::const_iterator iVar = boolvars.find(varTok->varId());
             if (iVar != boolvars.end() && iVar->second && // Variable has to be a boolean
                 ((tok->tokAt(1)->str() != "==" && tok->tokAt(1)->str() != "!=") ||
-                 ((MathLib::toLongNumber(numTok->str()) != 0) && (!code_is_c() || MathLib::toLongNumber(numTok->str()) != 1)))) { // == 0 and != 0 are allowed, for C also == 1 and != 1
+                 ((MathLib::toLongNumber(numTok->str()) != 0) && (!_tokenizer->code_is_c() || MathLib::toLongNumber(numTok->str()) != 1)))) { // == 0 and != 0 are allowed, for C also == 1 and != 1
                 comparisonOfBoolWithIntError(varTok, numTok->str());
             }
         } else if (Token::Match(tok, "%num% >|>=|==|!=|<=|< %var%")) { // Comparing number with variable
@@ -1172,7 +1172,7 @@ void CheckOther::checkComparisonOfBoolWithInt()
             std::map<unsigned int, bool>::const_iterator iVar = boolvars.find(varTok->varId());
             if (iVar != boolvars.end() && iVar->second && // Variable has to be a boolean
                 ((tok->tokAt(1)->str() != "==" && tok->tokAt(1)->str() != "!=") ||
-                 ((MathLib::toLongNumber(numTok->str()) != 0) && (!code_is_c() || MathLib::toLongNumber(numTok->str()) != 1)))) { // == 0 and != 0 are allowed, for C also == 1 and != 1
+                 ((MathLib::toLongNumber(numTok->str()) != 0) && (!_tokenizer->code_is_c() || MathLib::toLongNumber(numTok->str()) != 1)))) { // == 0 and != 0 are allowed, for C also == 1 and != 1
                 comparisonOfBoolWithIntError(varTok, numTok->str());
             }
         } else if (Token::Match(tok, "true|false >|>=|==|!=|<=|< %var%")) { // Comparing boolean constant with variable
@@ -1899,24 +1899,10 @@ static bool isFunction(const std::string &name, const Token *startToken)
     return false;
 }
 
-bool CheckOther::code_is_c() const
-{
-    const std::string fname = _tokenizer->getFiles()->at(0);
-    const size_t position   = fname.rfind(".");
-
-    if (position != std::string::npos) {
-        const std::string ext = fname.substr(position);
-        if (ext == ".c" || ext == ".C")
-            return true;
-    }
-
-    return false;
-}
-
 void CheckOther::checkMisusedScopedObject()
 {
     // Skip this check for .c files
-    if (code_is_c()) {
+    if (_tokenizer->code_is_c()) {
         return;
     }
 

lib/checkother.h

@@ -398,8 +398,6 @@ private:
 
         return varname;
     }
-
-    bool code_is_c() const;
 };
 /// @}
 //---------------------------------------------------------------------------

lib/tokenize.cpp

@@ -2338,6 +2338,9 @@ bool Tokenizer::tokenize(std::istream &code,
     // operator = => operator=
     simplifyOperatorName();
 
+    // Simplify pointer to standard types (C only)
+    simplifyPointerToStandardType();
+
     // simplify function pointers
     simplifyFunctionPointers();
 
@@ -5390,6 +5393,22 @@ void Tokenizer::simplifyFunctionParameters()
     }
 }
 
+void Tokenizer::simplifyPointerToStandardType()
+{
+    if (!code_is_c())
+        return;
+
+    for (Token *tok = _tokens; tok; tok = tok->next()) {
+        if (!Token::Match(tok, "& %var% [ 0 ]"))
+            continue;
+
+        // Remove '[ 0 ]' suffix
+        tok->next()->eraseTokens(tok->next(), tok->tokAt(5));
+        // Remove '&' prefix
+        tok = tok->previous();
+        tok->deleteNext();
+    }
+}
 
 void Tokenizer:: simplifyFunctionPointers()
 {
@@ -9599,3 +9618,17 @@ void Tokenizer::printUnknownTypes()
             _errorLogger->reportOut(ss.str());
     }
 }
+
+bool Tokenizer::code_is_c() const
+{
+    const std::string fname = getFiles()->at(0);
+    const size_t position   = fname.rfind(".");
+
+    if (position != std::string::npos) {
+        const std::string ext = fname.substr(position);
+        if (ext == ".c" || ext == ".C")
+            return true;
+    }
+
+    return false;
+}

lib/tokenize.h

@@ -514,6 +514,9 @@ public:
      */
     void simplifyStd();
 
+    /** Simplify pointer to standard type (C only) */
+    void simplifyPointerToStandardType();
+
     /** Simplify function pointers */
     void simplifyFunctionPointers();
 
@@ -711,6 +714,9 @@ public:
      */
     void printUnknownTypes();
 
+    /** Checks if the file extensions is .c or .C */
+    bool code_is_c() const;
+
 private:
     /** Disable copy constructor, no implementation */
     Tokenizer(const Tokenizer &);

test/testbufferoverrun.cpp

@@ -35,7 +35,7 @@ private:
 
 
 
-    void check(const char code[], bool experimental = true) {
+    void check(const char code[], bool experimental = true, const std::string &filename="test.cpp") {
         // Clear the error buffer..
         errout.str("");
 
@@ -49,7 +49,7 @@ private:
         // Tokenize..
         Tokenizer tokenizer(&settings, this);
         std::istringstream istr(code);
-        tokenizer.tokenize(istr, "test.cpp");
+        tokenizer.tokenize(istr, filename.c_str());
 
         // Assign variable ids
         tokenizer.simplifyTokenList();
@@ -3436,6 +3436,15 @@ private:
               "}\n");
         ASSERT_EQUALS("[test.cpp:4]: (warning) The buffer 'buf' is not zero-terminated after the call to readlink().\n", errout.str());
 
+        // C only: Primitive pointer simplification
+        check("void f()\n"
+              "{\n"
+              "    char buf[255];\n"
+              "    ssize_t len = readlink(path, &buf[0], sizeof(buf)-1);\n"
+              "    printf(\"%s\n\", buf);\n"
+              "}\n", true, "test.c");
+        ASSERT_EQUALS("[test.c:4]: (warning) The buffer 'buf' is not zero-terminated after the call to readlink().\n", errout.str());
+
         check("void f()\n"
               "{\n"
               "    char buf[255];\n"

test/testtokenize.cpp

@@ -286,6 +286,7 @@ private:
         TEST_CASE(simplifyConst);
         TEST_CASE(switchCase);
 
+        TEST_CASE(simplifyPointerToStandardType);
         TEST_CASE(functionpointer1);
         TEST_CASE(functionpointer2);
         TEST_CASE(functionpointer3);
@@ -375,7 +376,7 @@ private:
     }
 
 
-    std::string tokenizeAndStringify(const char code[], bool simplify = false, bool expand = true, Settings::PlatformType platform = Settings::Unspecified) {
+    std::string tokenizeAndStringify(const char code[], bool simplify = false, bool expand = true, Settings::PlatformType platform = Settings::Unspecified, const std::string &filename="test.cpp") {
         errout.str("");
 
         Settings settings;
@@ -385,7 +386,7 @@ private:
         // tokenize..
         Tokenizer tokenizer(&settings, this);
         std::istringstream istr(code);
-        tokenizer.tokenize(istr, "test.cpp");
+        tokenizer.tokenize(istr, filename.c_str());
         if (simplify)
             tokenizer.simplifyTokenList();
 
@@ -4669,6 +4670,21 @@ private:
                       tokenizeAndStringify("void foo (int i) { switch(i) { case -1: break; } }"));
     }
 
+    void simplifyPointerToStandardType() {
+        // Pointer to standard type
+        ASSERT_EQUALS("char buf [ 100 ] ; readlink ( path , buf , 99 ) ;",
+                      tokenizeAndStringify("char buf[100] ; readlink(path, &buf[0], 99);",
+                                           false, true, Settings::Unspecified, "test.c"));
+
+        // Simplification of unknown type - C only
+        ASSERT_EQUALS("foo data [ 100 ] ; something ( foo ) ;",
+                      tokenizeAndStringify("foo data[100]; something(&foo[0]);", false, true, Settings::Unspecified, "test.c"));
+
+        // C++: No pointer simplification
+        ASSERT_EQUALS("foo data [ 100 ] ; something ( & foo [ 0 ] ) ;",
+                      tokenizeAndStringify("foo data[100]; something(&foo[0]);"));
+    }
+
     std::string simplifyFunctionPointers(const char code[]) {
         errout.str("");
         Settings settings;