Move lang-code checking for case-mapping to C++ from JS

src/runtime/runtime-i18n.cc

Index: src/runtime/runtime-i18n.cc
diff --git a/src/runtime/runtime-i18n.cc b/src/runtime/runtime-i18n.cc
index 2d0c55eb217b1c328e09633d763f9c8d81c7bf8e..ce2650727a24d48c12aa612ffdd9c7c5d09937cf 100644
--- a/src/runtime/runtime-i18n.cc
+++ b/src/runtime/runtime-i18n.cc
@@ -1010,21 +1010,14 @@ inline int FindFirstUpperOrNonAscii(Handle<String> s, int length) {
   return length;
 }
 
-}  // namespace
-
-RUNTIME_FUNCTION(Runtime_StringToLowerCaseI18N) {
-  HandleScope scope(isolate);
-  DCHECK_EQ(args.length(), 1);
-  CONVERT_ARG_HANDLE_CHECKED(String, s, 0);
-
-  int length = s->length();
-  s = String::Flatten(s);
-
+MUST_USE_RESULT Object* ConvertToLower(Handle<String> s, Isolate* isolate) {
   if (!s->HasOnlyOneByteChars()) {
     // Use a slower implementation for strings with characters beyond U+00FF.
     return LocaleConvertCase(s, isolate, false, "");
   }
 
+  int length = s->length();
+
   // We depend here on the invariant that the length of a Latin1
   // string is invariant under ToLowerCase, and the result always
   // fits in the Latin1 range in the *root locale*. It does not hold
@@ -1080,14 +1073,8 @@ RUNTIME_FUNCTION(Runtime_StringToLowerCaseI18N) {
   return *result;
 }
 
-RUNTIME_FUNCTION(Runtime_StringToUpperCaseI18N) {
-  HandleScope scope(isolate);
-  DCHECK_EQ(args.length(), 1);
-  CONVERT_ARG_HANDLE_CHECKED(String, s, 0);
-
+MUST_USE_RESULT Object* ConvertToUpper(Handle<String> s, Isolate* isolate) {
   int32_t length = s->length();
-  s = String::Flatten(s);
-
   if (s->HasOnlyOneByteChars() && length > 0) {
     Handle<SeqOneByteString> result =
         isolate->factory()->NewRawOneByteString(length).ToHandleChecked();
@@ -1147,26 +1134,65 @@ RUNTIME_FUNCTION(Runtime_StringToUpperCaseI18N) {
   return LocaleConvertCase(s, isolate, true, "");
 }
 
+MUST_USE_RESULT Object* ConvertCase(Handle<String> s, bool is_upper,
+                                    Isolate* isolate) {
+  return is_upper ? ConvertToUpper(s, isolate) : ConvertToLower(s, isolate);
+}
+
+}  // namespace
+
+RUNTIME_FUNCTION(Runtime_StringToLowerCaseI18N) {
+  HandleScope scope(isolate);
+  DCHECK_EQ(args.length(), 1);
+  CONVERT_ARG_HANDLE_CHECKED(String, s, 0);
+  s = String::Flatten(s);
+  return ConvertToLower(s, isolate);
+}
+
+RUNTIME_FUNCTION(Runtime_StringToUpperCaseI18N) {
+  HandleScope scope(isolate);
+  DCHECK_EQ(args.length(), 1);
+  CONVERT_ARG_HANDLE_CHECKED(String, s, 0);
+  s = String::Flatten(s);
+  return ConvertToUpper(s, isolate);
+}
+
 RUNTIME_FUNCTION(Runtime_StringLocaleConvertCase) {
   HandleScope scope(isolate);
   DCHECK_EQ(args.length(), 3);
   CONVERT_ARG_HANDLE_CHECKED(String, s, 0);
   CONVERT_BOOLEAN_ARG_CHECKED(is_upper, 1);
-  CONVERT_ARG_HANDLE_CHECKED(SeqOneByteString, lang, 2);

[jungshik at Google, 2017/01/11 19:11:22]

From i18n.js, a string "literal" ( e.g. "el", "tr", "az", "lt" ) is passed and I
assumed that it'd be SeqOneByteString. However, my analysis of minidump from
Windows Chrome canary builds on Windows indicates that the above check fails. 

So, in this CL, I'm not relying on that assumption any more by moving language
code check from i18n.js to runtime-i18n.cc. 

-
-  // All the languages requiring special handling ("az", "el", "lt", "tr")
-  // have a 2-letter language code.
-  DCHECK(lang->length() == 2);
-  uint8_t lang_str[3];
-  memcpy(lang_str, lang->GetChars(), 2);
-  lang_str[2] = 0;
+  CONVERT_ARG_HANDLE_CHECKED(String, lang_arg, 2);
+
+  DCHECK(lang_arg->length() <= 3);
+  lang_arg = String::Flatten(lang_arg);
+
+  // All the languages requiring special-handling have two-letter codes.
+  if (V8_UNLIKELY(lang_arg->length() > 2))
+    return ConvertCase(s, is_upper, isolate);
+
+  char c1, c2;
+  {
+    DisallowHeapAllocation no_gc;
+    String::FlatContent lang = lang_arg->GetFlatContent();
+    c1 = lang.Get(0);
+    c2 = lang.Get(1);
+  }
   s = String::Flatten(s);
   // TODO(jshin): Consider adding a fast path for ASCII or Latin-1. The fastpath
   // in the root locale needs to be adjusted for az, lt and tr because even case
   // mapping of ASCII range characters are different in those locales.
-  // Greek (el) does not require any adjustment, though.
-  return LocaleConvertCase(s, isolate, is_upper,
-                           reinterpret_cast<const char*>(lang_str));
+  // Greek (el) does not require any adjustment.
+  if (V8_UNLIKELY(c1 == 't' && c2 == 'r'))
+    return LocaleConvertCase(s, isolate, is_upper, "tr");
+  if (V8_UNLIKELY(c1 == 'e' && c2 == 'l'))
+    return LocaleConvertCase(s, isolate, is_upper, "el");
+  if (V8_UNLIKELY(c1 == 'l' && c2 == 't'))
+    return LocaleConvertCase(s, isolate, is_upper, "lt");
+  if (V8_UNLIKELY(c1 == 'a' && c2 == 'z'))
+    return LocaleConvertCase(s, isolate, is_upper, "az");
+
+  return ConvertCase(s, is_upper, isolate);
 }
 
 RUNTIME_FUNCTION(Runtime_DateCacheVersion) {