upstream cl 182543808

1 files changed, 84 insertions, 4 deletions

diff --git a/googletest/src/gtest-printers.cc b/googletest/src/gtest-printers.cc
index dd67f64..1bdf243 100644
--- a/googletest/src/gtest-printers.cc
+++ b/googletest/src/gtest-printers.cc
@@ -43,12 +43,13 @@
 // defines Foo.
 
 #include "gtest/gtest-printers.h"
-#include <ctype.h>
 #include <stdio.h>
+#include <cctype>
 #include <cwchar>
 #include <ostream>  // NOLINT
 #include <string>
 #include "gtest/internal/gtest-port.h"
+#include "src/gtest-internal-inl.h"
 
 namespace testing {
 
@@ -262,11 +263,12 @@ template <typename CharType>
 GTEST_ATTRIBUTE_NO_SANITIZE_MEMORY_
 GTEST_ATTRIBUTE_NO_SANITIZE_ADDRESS_
 GTEST_ATTRIBUTE_NO_SANITIZE_THREAD_
-static void PrintCharsAsStringTo(
+static CharFormat PrintCharsAsStringTo(
     const CharType* begin, size_t len, ostream* os) {
   const char* const kQuoteBegin = sizeof(CharType) == 1 ? "\"" : "L\"";
   *os << kQuoteBegin;
   bool is_previous_hex = false;
+  CharFormat print_format = kAsIs;
   for (size_t index = 0; index < len; ++index) {
     const CharType cur = begin[index];
     if (is_previous_hex && IsXDigit(cur)) {
@@ -276,8 +278,13 @@ static void PrintCharsAsStringTo(
       *os << "\" " << kQuoteBegin;
     }
     is_previous_hex = PrintAsStringLiteralTo(cur, os) == kHexEscape;
+    // Remember if any characters required hex escaping.
+    if (is_previous_hex) {
+      print_format = kHexEscape;
+    }
   }
   *os << "\"";
+  return print_format;
 }
 
 // Prints a (const) char/wchar_t array of 'len' elements, starting at address
@@ -347,15 +354,88 @@ void PrintTo(const wchar_t* s, ostream* os) {
 }
 #endif  // wchar_t is native
 
+namespace {
+
+bool ContainsUnprintableControlCodes(const char* str, size_t length) {
+  for (size_t i = 0; i < length; i++) {
+    char ch = *str++;
+    if (std::iscntrl(ch)) {
+        switch (ch) {
+        case '\t':
+        case '\n':
+        case '\r':
+          break;
+        default:
+          return true;
+        }
+      }
+  }
+  return false;
+}
+
+bool IsUTF8TrailByte(unsigned char t) { return 0x80 <= t && t<= 0xbf; }
+
+bool IsValidUTF8(const char* str, size_t length) {
+  const unsigned char *s = reinterpret_cast<const unsigned char *>(str);
+
+  for (size_t i = 0; i < length;) {
+    unsigned char lead = s[i++];
+
+    if (lead <= 0x7f) {
+      continue;  // single-byte character (ASCII) 0..7F
+    }
+    if (lead < 0xc2) {
+      return false;  // trail byte or non-shortest form
+    } else if (lead <= 0xdf && (i + 1) <= length && IsUTF8TrailByte(s[i])) {
+      ++i;  // 2-byte character
+    } else if (0xe0 <= lead && lead <= 0xef && (i + 2) <= length &&
+               IsUTF8TrailByte(s[i]) &&
+               IsUTF8TrailByte(s[i + 1]) &&
+               // check for non-shortest form and surrogate
+               (lead != 0xe0 || s[i] >= 0xa0) &&
+               (lead != 0xed || s[i] < 0xa0)) {
+      i += 2;  // 3-byte character
+    } else if (0xf0 <= lead && lead <= 0xf4 && (i + 3) <= length &&
+               IsUTF8TrailByte(s[i]) &&
+               IsUTF8TrailByte(s[i + 1]) &&
+               IsUTF8TrailByte(s[i + 2]) &&
+               // check for non-shortest form
+               (lead != 0xf0 || s[i] >= 0x90) &&
+               (lead != 0xf4 || s[i] < 0x90)) {
+      i += 3;  // 4-byte character
+    } else {
+      return false;
+    }
+  }
+  return true;
+}
+
+void ConditionalPrintAsText(const char* str, size_t length, ostream* os) {
+  if (!ContainsUnprintableControlCodes(str, length) &&
+      IsValidUTF8(str, length)) {
+    *os << "\n    As Text: \"" << str << "\"";
+  }
+}
+
+}  // anonymous namespace
+
 // Prints a ::string object.
 #if GTEST_HAS_GLOBAL_STRING
 void PrintStringTo(const ::string& s, ostream* os) {
-  PrintCharsAsStringTo(s.data(), s.size(), os);
+  if (PrintCharsAsStringTo(s.data(), s.size(), os) == kHexEscape) {
+    if (GTEST_FLAG(print_utf8)) {
+      ConditionalPrintAsText(s.data(), s.size(), os);
+    }
+  }
 }
 #endif  // GTEST_HAS_GLOBAL_STRING
 
 void PrintStringTo(const ::std::string& s, ostream* os) {
-  PrintCharsAsStringTo(s.data(), s.size(), os);
+  if (PrintCharsAsStringTo(s.data(), s.size(), os) == kHexEscape) {
+    if (GTEST_FLAG(print_utf8)) {
+      ConditionalPrintAsText(s.data(), s.size(), os);
+    }
+  }
 }
 
 // Prints a ::wstring object.