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

src/runtime/runtime-i18n.cc

 // Copyright 2014 the V8 project authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
 
 #ifdef V8_I18N_SUPPORT
 #include "src/runtime/runtime-utils.h"
 
 #include <memory>
 
@@ skipping 992 matching lines @@
 inline int FindFirstUpperOrNonAscii(Handle<String> s, int length) {
   for (int index = 0; index < length; ++index) {
     uint16_t ch = s->Get(index);
     if (V8_UNLIKELY(IsASCIIUpper(ch) || ch & ~0x7F)) {
       return index;
     }
   }
   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
   // for ToUpperCase even in the root locale.
 
   // Scan the string for uppercase and non-ASCII characters for strings
   // shorter than a machine-word without any memory allocation overhead.
   // TODO(jshin): Apply this to a longer input by breaking FastAsciiConvert()
   // to two parts, one for scanning the prefix with no change and the other for
   // handling ASCII-only characters.
@@ skipping 35 matching lines @@
     const uint16_t* src = flat.ToUC16Vector().start();
     CopyChars(dest, src, index_to_first_unprocessed);
     for (int index = index_to_first_unprocessed; index < length; ++index) {
       dest[index] = ToLatin1Lower(static_cast<uint16_t>(src[index]));
     }
   }
 
   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();
 
     int sharp_s_count;
     bool is_result_single_byte;
     {
       DisallowHeapAllocation no_gc;
       String::FlatContent flat = s->GetFlatContent();
       uint8_t* dest = result->GetChars();
@@ skipping 39 matching lines @@
     } else {
       ToUpperWithSharpS(flat.ToUC16Vector(), result);
     }
 
     return *result;
   }
 
   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. 

+  CONVERT_ARG_HANDLE_CHECKED(String, lang_arg, 2);
 
-  // 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;
+  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) {
   HandleScope scope(isolate);
   DCHECK_EQ(0, args.length());
   if (isolate->serializer_enabled()) return isolate->heap()->undefined_value();
   if (!isolate->eternal_handles()->Exists(EternalHandles::DATE_CACHE_VERSION)) {
     Handle<FixedArray> date_cache_version =
         isolate->factory()->NewFixedArray(1, TENURED);
     date_cache_version->set(0, Smi::kZero);
     isolate->eternal_handles()->CreateSingleton(
         isolate, *date_cache_version, EternalHandles::DATE_CACHE_VERSION);
   }
   Handle<FixedArray> date_cache_version =
       Handle<FixedArray>::cast(isolate->eternal_handles()->GetSingleton(
           EternalHandles::DATE_CACHE_VERSION));
   return date_cache_version->get(0);
 }
 
 }  // namespace internal
 }  // namespace v8
 
 #endif  // V8_I18N_SUPPORT