Fixed #4481 (Simplify %str% [ %num% ])

lib/token.cpp

@@ -741,6 +741,34 @@ std::size_t Token::getStrLength(const Token *tok)
     return len;
 }
 
+std::string Token::getCharAt(const Token *tok, std::size_t index)
+{
+    assert(tok != NULL);
+
+    const std::string strValue(tok->strValue());
+    const char *str = strValue.c_str();
+
+    while (*str) {
+        if (index == 0) {
+            std::string ret;
+            if (*str == '\\') {
+                ret = *str;
+                ++str;
+            }
+            ret += *str;
+            return ret;
+        }
+
+        if (*str == '\\')
+            ++str;
+        ++str;
+        --index;
+    }
+    assert(index == 0);
+
+    return "\\0";
+}
+
 void Token::move(Token *srcStart, Token *srcEnd, Token *newLocation)
 {
     /**[newLocation] -> b -> c -> [srcStart] -> [srcEnd] -> f */

lib/token.h

@@ -178,6 +178,16 @@ public:
      **/
     static std::size_t getStrLength(const Token *tok);
 
+    /**
+     * Return char of C-string at index (possible escaped "\\n")
+     *
+     * Should be called for %str% tokens only.
+     *
+     * @param tok token with C-string
+     * @param index position of character
+     **/
+    static std::string getCharAt(const Token *tok, std::size_t index);
+
     Type type() const {
         return _type;
     }

lib/tokenize.cpp

@@ -3244,6 +3244,8 @@ bool Tokenizer::simplifyTokenList()
     // clear the _functionList so it can't contain dead pointers
     deleteSymbolDatabase();
 
+    simplifyCharAt();
+
     // simplify references
     simplifyReference();
 
@@ -6661,6 +6663,21 @@ bool Tokenizer::simplifyRedundantParenthesis()
     return ret;
 }
 
+void Tokenizer::simplifyCharAt()
+{
+    // Replace "string"[0] with 's'
+    for (Token *tok = list.front(); tok; tok = tok->next()) {
+        if (Token::Match(tok, "%str% [ %num% ]")) {
+            const MathLib::bigint index = MathLib::toLongNumber(tok->tokAt(2)->str());
+            // Check within range
+            if (index >= 0 && index <= (MathLib::bigint)Token::getStrLength(tok)) {
+                tok->str(std::string("'" + Token::getCharAt(tok, (size_t)index) + "'"));
+                tok->deleteNext(3);
+            }
+        }
+    }
+}
+
 void Tokenizer::simplifyReference()
 {
     if (isC())

lib/tokenize.h

@@ -400,6 +400,8 @@ public:
      */
     bool simplifyRedundantParenthesis();
 
+    void simplifyCharAt();
+
     /** Simplify references */
     void simplifyReference();
 

test/testsimplifytokens.cpp

@@ -433,6 +433,7 @@ private:
         TEST_CASE(undefinedSizeArray);
 
         TEST_CASE(simplifyArrayAddress);  // Replace "&str[num]" => "(str + num)"
+        TEST_CASE(simplifyCharAt);
     }
 
     std::string tok(const char code[], bool simplify = true, Settings::PlatformType type = Settings::Unspecified) {
@@ -8046,6 +8047,18 @@ private:
         // Don't crash
         tok("int", true);
     }
+
+    void simplifyCharAt() { // ticket #4481
+        ASSERT_EQUALS("'h' ;", tok("\"hello\"[0] ;"));
+        ASSERT_EQUALS("'\n' ;", tok("\"\n\"[0] ;"));
+        ASSERT_EQUALS("'\\0' ;", tok("\"hello\"[5] ;"));
+        ASSERT_EQUALS("'\\0' ;", tok("\"\"[0] ;"));
+        ASSERT_EQUALS("'\\0' ;", tok("\"\\0\"[0] ;"));
+        ASSERT_EQUALS("'\\n' ;", tok("\"hello\\nworld\"[5] ;"));
+        ASSERT_EQUALS("'w' ;", tok("\"hello\nworld\"[6] ;"));
+        ASSERT_EQUALS("\"hello\" [ 7 ] ;", tok("\"hello\"[7] ;"));
+        ASSERT_EQUALS("\"hello\" [ -1 ] ;", tok("\"hello\"[-1] ;"));
+    }
 };
 
 REGISTER_TEST(TestSimplifyTokens)