Fix #714 (# operator escapes backslashes outside string literals)

Author: dmcppcheck

run-tests.py

@@ -80,7 +80,6 @@ def cleanup(out: str) -> str:
          'macro_rescan_varargs.c',
 
          # todo, high priority
-         'c99-6_10_3_4_p6.c',
          'expr_usual_conversions.c', # condition is true: 4U - 30 >= 0
          'stdint.c',
 

simplecpp.cpp

@@ -642,6 +642,23 @@ static bool isStringLiteralPrefix(const std::string &str)
            str == "R" || str == "uR" || str == "UR" || str == "LR" || str == "u8R";
 }
 
+/** Spelling of a token in the result of the # operator: a \ is inserted
+ *  before each " and \ of a string literal or character constant (C 6.10.3.2p2) */
+static std::string stringifyToken(const std::string &str)
+{
+    const std::string::size_type quote = str.find_first_of("\"\'");
+    if (quote == std::string::npos || str.back() != str[quote] ||
+        (quote > 0 && !isStringLiteralPrefix(str.substr(0, quote))))
+        return str;
+    std::string ret;
+    for (const char c : str) {
+        if (c == '\\' || c == '\"')
+            ret += '\\';
+        ret += c;
+    }
+    return ret;
+}
+
 void simplecpp::TokenList::lineDirective(unsigned int fileIndex_, unsigned int line, Location &location)
 {
     if (fileIndex_ != location.fileIndex || line >= location.line) {
@@ -2381,12 +2398,12 @@ namespace simplecpp {
             ostr << '\"';
             for (const Token *hashtok = tokenListHash.cfront(), *next; hashtok; hashtok = next) {
                 next = hashtok->next;
-                ostr << hashtok->str();
+                ostr << stringifyToken(hashtok->str());
                 if (next && hashtok->whitespaceahead)
                     ostr << ' ';
             }
             ostr << '\"';
-            output.push_back(newMacroToken(escapeString(ostr.str()), loc, isReplaced(expandedmacros)));
+            output.push_back(newMacroToken(ostr.str(), loc, isReplaced(expandedmacros)));
             return tok;
         }
 

test.cpp

@@ -1709,6 +1709,25 @@ static void hash()
                              "STR(X(Y,Y))"));
 }
 
+static void hashEscape()
+{
+    // a \ is only inserted before " and \ inside string literals and character constants
+    const char code[] = "#define str(x) #x\n"
+                        "str(\\n)\n"
+                        "str(: @\\n)\n"
+                        "str(\"\\n\")\n"
+                        "str('\\4')\n"
+                        "str('\"')\n"
+                        "str(L\"a\\\"b\")";
+    ASSERT_EQUALS("\n"
+                  "\"\\n\"\n"
+                  "\": @\\n\"\n"
+                  "\"\\\"\\\\n\\\"\"\n"
+                  "\"'\\\\4'\"\n"
+                  "\"'\\\"'\"\n"
+                  "\"L\\\"a\\\\\\\"b\\\"\"", preprocess(code));
+}
+
 static void hashhash1()   // #4703
 {
     const char code[] = "#define MACRO( A, B, C ) class A##B##C##Creator {};\n"
@@ -4734,6 +4753,7 @@ static void runTests(int argc, char **argv, Input input)
     TEST_CASE(garbage_endif);
 
     TEST_CASE(hash);
+    TEST_CASE(hashEscape);
     TEST_CASE(hashhash1);
     TEST_CASE(hashhash2);
     TEST_CASE(hashhash3);