compare special-value symbols against the whole input character

double-conversion/string-to-double.cc

@@ -51,13 +51,25 @@ namespace double_conversion {
 
 namespace {
 
-inline char ToLower(char ch) {
+// Widens an input character to its unsigned code-unit value. Symbol matching
+// compares characters in this form, so that a 16-bit input character is never
+// truncated into the range of the symbol byte it is compared against.
+inline uint32_t CodeUnit(char ch) {
+  return static_cast<unsigned char>(ch);
+}
+
+inline uint32_t CodeUnit(uc16 ch) {
+  return ch;
+}
+
+inline uint32_t ToLower(uint32_t ch) {
+  if (ch > 0x7F) return ch;
   static const std::ctype<char>& cType =
       std::use_facet<std::ctype<char> >(std::locale::classic());
-  return cType.tolower(ch);
+  return static_cast<unsigned char>(cType.tolower(static_cast<char>(ch)));
 }
 
-inline char Pass(char ch) {
+inline uint32_t Pass(uint32_t ch) {
   return ch;
 }
 
@@ -66,10 +78,11 @@ static inline bool ConsumeSubStringImpl(Iterator* current,
                                         Iterator end,
                                         const char* substring,
                                         Converter converter) {
-  DOUBLE_CONVERSION_ASSERT(converter(**current) == *substring);
+  DOUBLE_CONVERSION_ASSERT(converter(CodeUnit(**current)) == CodeUnit(*substring));
   for (substring++; *substring != '\0'; substring++) {
     ++*current;
-    if (*current == end || converter(**current) != *substring) {
+    if (*current == end ||
+        converter(CodeUnit(**current)) != CodeUnit(*substring)) {
       return false;
     }
   }
@@ -92,10 +105,13 @@ static bool ConsumeSubString(Iterator* current,
 }
 
 // Consumes first character of the str is equal to ch
-inline bool ConsumeFirstCharacter(char ch,
+template <class Char>
+inline bool ConsumeFirstCharacter(Char ch,
                                          const char* str,
                                          bool case_insensitivity) {
-  return case_insensitivity ? ToLower(ch) == str[0] : ch == str[0];
+  const uint32_t c = CodeUnit(ch);
+  const uint32_t first = CodeUnit(str[0]);
+  return case_insensitivity ? ToLower(c) == first : c == first;
 }
 }  // namespace
 

test/cctest/test-conversions.cc

@@ -6002,6 +6002,60 @@ TEST(StringToDoubleCaseInsensitiveSpecialValues) {
 }
 
 
+TEST(StringToDoubleNonAsciiSpecialValues) {
+  int processed = 0;
+
+  // Use 1.0 as junk_string_value.
+  StringToDoubleConverter converter(StringToDoubleConverter::NO_FLAGS,
+                                    0.0, 1.0, "Infinity", "NaN");
+
+  // The ASCII spellings are the only ones that match.
+  const uc16 nan16[] = { 'N', 'a', 'N' };
+  CHECK_EQ(Double::NaN(), converter.StringToDouble(nan16, 3, &processed));
+  CHECK_EQ(3, processed);
+
+  const uc16 infinity16[] = { 'I', 'n', 'f', 'i', 'n', 'i', 't', 'y' };
+  CHECK_EQ(Double::Infinity(),
+           converter.StringToDouble(infinity16, 8, &processed));
+  CHECK_EQ(8, processed);
+
+  // Characters whose low byte equals a symbol character must not match it.
+  // U+014E and U+FF4E end in 'N', U+0161 ends in 'a', U+FF49 ends in 'I' and
+  // U+016E ends in 'n'.
+  const uc16 nan_lead[] = { 0x014E, 'a', 'N' };
+  CHECK_EQ(1.0, converter.StringToDouble(nan_lead, 3, &processed));
+  CHECK_EQ(0, processed);
+
+  const uc16 nan_tail[] = { 'N', 0x0161, 'N' };
+  CHECK_EQ(1.0, converter.StringToDouble(nan_tail, 3, &processed));
+  CHECK_EQ(0, processed);
+
+  const uc16 infinity_lead[] = { 0xFF49, 'n', 'f', 'i', 'n', 'i', 't', 'y' };
+  CHECK_EQ(1.0, converter.StringToDouble(infinity_lead, 8, &processed));
+  CHECK_EQ(0, processed);
+
+  const uc16 infinity_tail[] = { 'I', 0x016E, 'f', 'i', 'n', 'i', 't', 'y' };
+  CHECK_EQ(1.0, converter.StringToDouble(infinity_tail, 8, &processed));
+  CHECK_EQ(0, processed);
+
+  // The same holds when the symbols are matched case-insensitively.
+  const int ci_flags = StringToDoubleConverter::ALLOW_CASE_INSENSITIVITY;
+  StringToDoubleConverter ci_converter(ci_flags, 0.0, 1.0, "infinity", "nan");
+
+  const uc16 nan_ci[] = { 'n', 'a', 'N' };
+  CHECK_EQ(Double::NaN(), ci_converter.StringToDouble(nan_ci, 3, &processed));
+  CHECK_EQ(3, processed);
+
+  const uc16 nan_ci_lead[] = { 0x014E, 'a', 'n' };
+  CHECK_EQ(1.0, ci_converter.StringToDouble(nan_ci_lead, 3, &processed));
+  CHECK_EQ(0, processed);
+
+  const uc16 nan_ci_tail[] = { 'n', 'a', 0x014E };
+  CHECK_EQ(1.0, ci_converter.StringToDouble(nan_ci_tail, 3, &processed));
+  CHECK_EQ(0, processed);
+}
+
+
 TEST(StringToTemplate) {
     // Test StringToDoubleConverter::StringTo<T>.