Makes gtest print string literals correctly when it contains \x escape sequences. Contributed by Yair Chuchem.

git-svn-id: http://googletest.googlecode.com/svn/trunk@525 861a406c-534a-0410-8894-cb66d6ee9925

3 files changed, 56 insertions, 12 deletions

diff --git a/include/gtest/internal/gtest-port.h b/include/gtest/internal/gtest-port.h
index f08f6df..900a41d 100644
--- a/include/gtest/internal/gtest-port.h
+++ b/include/gtest/internal/gtest-port.h
@@ -1462,6 +1462,9 @@ inline bool IsSpace(char ch) {
 inline bool IsUpper(char ch) {
   return isupper(static_cast<unsigned char>(ch)) != 0;
 }
+inline bool IsXDigit(char ch) {
+  return isxdigit(static_cast<unsigned char>(ch)) != 0;
+}
 
 inline char ToLower(char ch) {
   return static_cast<char>(tolower(static_cast<unsigned char>(ch)));
diff --git a/src/gtest-printers.cc b/src/gtest-printers.cc
index 147f8b2..c85d582 100644
--- a/src/gtest-printers.cc
+++ b/src/gtest-printers.cc
@@ -194,25 +194,25 @@ static CharFormat PrintAsCharLiteralTo(Char c, ostream* os) {
   return kSpecialEscape;
 }
 
-// Prints a char as if it's part of a string literal, escaping it when
-// necessary.
-static void PrintAsWideStringLiteralTo(wchar_t c, ostream* os) {
+// Prints a char c as if it's part of a string literal, escaping it when
+// necessary; returns how c was formatted.
+static CharFormat PrintAsWideStringLiteralTo(wchar_t c, ostream* os) {
   switch (c) {
     case L'\'':
       *os << "'";
-      break;
+      return kAsIs;
     case L'"':
       *os << "\\\"";
-      break;
+      return kSpecialEscape;
     default:
-      PrintAsCharLiteralTo<wchar_t>(c, os);
+      return PrintAsCharLiteralTo<wchar_t>(c, os);
   }
 }
 
-// Prints a char as if it's part of a string literal, escaping it when
-// necessary.
-static void PrintAsNarrowStringLiteralTo(char c, ostream* os) {
-  PrintAsWideStringLiteralTo(static_cast<unsigned char>(c), os);
+// Prints a char c as if it's part of a string literal, escaping it when
+// necessary; returns how c was formatted.
+static CharFormat PrintAsNarrowStringLiteralTo(char c, ostream* os) {
+  return PrintAsWideStringLiteralTo(static_cast<unsigned char>(c), os);
 }
 
 // Prints a wide or narrow character c and its code.  '\0' is printed
@@ -263,8 +263,16 @@ void PrintTo(wchar_t wc, ostream* os) {
 // and may not be null-terminated.
 static void PrintCharsAsStringTo(const char* begin, size_t len, ostream* os) {
   *os << "\"";
+  bool is_previous_hex = false;
   for (size_t index = 0; index < len; ++index) {
-    PrintAsNarrowStringLiteralTo(begin[index], os);
+    const char cur = begin[index];
+    if (is_previous_hex && IsXDigit(cur)) {
+      // Previous character is of '\x..' form and this character can be
+      // interpreted as another hexadecimal digit in its number. Break string to
+      // disambiguate.
+      *os << "\" \"";
+    }
+    is_previous_hex = PrintAsNarrowStringLiteralTo(cur, os) == kHexEscape;
   }
   *os << "\"";
 }
@@ -280,8 +288,17 @@ void UniversalPrintArray(const char* begin, size_t len, ostream* os) {
 static void PrintWideCharsAsStringTo(const wchar_t* begin, size_t len,
                                      ostream* os) {
   *os << "L\"";
+  bool is_previous_hex = false;
   for (size_t index = 0; index < len; ++index) {
-    PrintAsWideStringLiteralTo(begin[index], os);
+    const wchar_t cur = begin[index];
+    if (is_previous_hex && 0 <= cur && cur < 128 &&
+        IsXDigit(static_cast<char>(cur))) {
+      // Previous character is of '\x..' form and this character can be
+      // interpreted as another hexadecimal digit in its number. Break string to
+      // disambiguate.
+      *os << "\" L\"";
+    }
+    is_previous_hex = PrintAsWideStringLiteralTo(cur, os) == kHexEscape;
   }
   *os << "\"";
 }
diff --git a/test/gtest-printers_test.cc b/test/gtest-printers_test.cc
index 5eabd23..da1fbc2 100644
--- a/test/gtest-printers_test.cc
+++ b/test/gtest-printers_test.cc
@@ -658,6 +658,20 @@ TEST(PrintStringTest, StringInStdNamespace) {
             Print(str));
 }
 
+TEST(PrintStringTest, StringAmbiguousHex) {
+  // "\x6BANANA" is ambiguous, it can be interpreted as starting with either of:
+  // '\x6', '\x6B', or '\x6BA'.
+
+  // a hex escaping sequence following by a decimal digit
+  EXPECT_EQ("\"0\\x12\" \"3\"", Print(::std::string("0\x12" "3")));
+  // a hex escaping sequence following by a hex digit (lower-case)
+  EXPECT_EQ("\"mm\\x6\" \"bananas\"", Print(::std::string("mm\x6" "bananas")));
+  // a hex escaping sequence following by a hex digit (upper-case)
+  EXPECT_EQ("\"NOM\\x6\" \"BANANA\"", Print(::std::string("NOM\x6" "BANANA")));
+  // a hex escaping sequence following by a non-xdigit
+  EXPECT_EQ("\"!\\x5-!\"", Print(::std::string("!\x5-!")));
+}
+
 // Tests printing ::wstring and ::std::wstring.
 
 #if GTEST_HAS_GLOBAL_WSTRING
@@ -680,6 +694,16 @@ TEST(PrintWideStringTest, StringInStdNamespace) {
             "\\xD3\\x576\\x8D3\\xC74D a\\0\"",
             Print(str));
 }
+
+TEST(PrintWideStringTest, StringAmbiguousHex) {
+  // same for wide strings.
+  EXPECT_EQ("L\"0\\x12\" L\"3\"", Print(::std::wstring(L"0\x12" L"3")));
+  EXPECT_EQ("L\"mm\\x6\" L\"bananas\"",
+            Print(::std::wstring(L"mm\x6" L"bananas")));
+  EXPECT_EQ("L\"NOM\\x6\" L\"BANANA\"",
+            Print(::std::wstring(L"NOM\x6" L"BANANA")));
+  EXPECT_EQ("L\"!\\x5-!\"", Print(::std::wstring(L"!\x5-!")));
+}
 #endif  // GTEST_HAS_STD_WSTRING
 
 // Tests printing types that support generic streaming (i.e. streaming