Experimental CL on top of https://codereview.chromium.org/1812673005/

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 8f05809c50c7c87007fb27d2b18d93b6299edd60..f4a3394c4c4d2efabfe81235c485f2230735797e 100644
--- a/src/runtime/runtime-i18n.cc
+++ b/src/runtime/runtime-i18n.cc
@@ -753,13 +753,6 @@ RUNTIME_FUNCTION(Runtime_BreakIteratorBreakType) {
 }
 
 namespace {
-inline void LocaleConvertCaseHelper(icu::UnicodeString* s, bool is_to_upper,
-                                    const icu::Locale& locale) {
-  if (is_to_upper)
-    s->toUpper(locale);
-  else
-    s->toLower(locale);
-}
 void ConvertCaseWithTransliterator(icu::UnicodeString* input,
                                    const char* transliterator_id) {
   UErrorCode status = U_ZERO_ERROR;
@@ -778,43 +771,79 @@ MUST_USE_RESULT Object* LocaleConvertCase(Handle<String> s, Isolate* isolate,
   };
   RUNTIME_ASSERT(locale_id >= -1 &&
                  locale_id < static_cast<int>(arraysize(conversion_locales)));
-  int32_t length = s->length();
-  icu::UnicodeString converted;
-  {
-    DisallowHeapAllocation no_gc;
-    DCHECK(s->IsFlat());
-    String::FlatContent flat = s->GetFlatContent();
+  int32_t src_length = s->length();
+  const UChar* src = nullptr;
 
-    const UChar* src;
-    if (flat.IsOneByte()) {
-      base::SmartArrayPointer<uc16> sap = s->ToWideCString();
-      src = reinterpret_cast<const UChar*>(sap.get());
-      converted = icu::UnicodeString(src, length);
-    } else {
-      src = reinterpret_cast<const UChar*>(flat.ToUC16Vector().start());
-      converted = icu::UnicodeString(src, length);
-    }
+  base::SmartArrayPointer<uc16> sap;
+  if (s->IsOneByteRepresentationUnderneath()) {
+    sap = s->ToWideCString();
+    src = reinterpret_cast<const UChar*>(sap.get());
   }
 
-  if (locale_id == -1) {
-    LocaleConvertCaseHelper(&converted, is_to_upper, icu::Locale::getRoot());
-  } else if (V8_UNLIKELY(locale_id == 1 && is_to_upper)) {
-    // TODO(jshin): Once http://bugs.icu-project.org/trac/ticket/10582 is
-    // fixed, remove this special-casing for uppercasing in Greek(el) locale.
-    // This is ~500 times slower than using the case conversion API.
-    ConvertCaseWithTransliterator(&converted, "el-Upper");
-  } else {
-    LocaleConvertCaseHelper(&converted, is_to_upper,
-                            icu::Locale(conversion_locales[locale_id]));
+  // Greek (id == 1) uppercasing has to be done via transliteration.
+  // TODO(jshin): Drop this special-casing once ICU's regular case conversion
+  // API supports Greek uppercasing. See
+  // http://bugs.icu-project.org/trac/ticket/10582 .
+  // ICU's C API for transliteration is nasty and we just use C++ API.
+  if (V8_UNLIKELY(locale_id == 1 && is_to_upper)) {
+    icu::UnicodeString converted;
+    {
+      DisallowHeapAllocation no_gc;
+      String::FlatContent flat = s->GetFlatContent();
+      if (src == nullptr) {
+        DCHECK(flat.IsTwoByte());
+        src = reinterpret_cast<const UChar*>(flat.ToUC16Vector().start());
+      }
+      // Starts with the source string and will be replaced by the converted
+      // result.
+      converted.fastCopyFrom(icu::UnicodeString(false, src, src_length));
+      ConvertCaseWithTransliterator(&converted, "el-Upper");
+    }
+    Handle<String> result;
+    ASSIGN_RETURN_FAILURE_ON_EXCEPTION(
+        isolate, result,
+        isolate->factory()->NewStringFromTwoByte(Vector<const uint16_t>(
+            reinterpret_cast<const uint16_t*>(converted.getBuffer()),
+            converted.length())));
+    return *result;
   }
 
-  Handle<String> result;
-  ASSIGN_RETURN_FAILURE_ON_EXCEPTION(
-      isolate, result,
-      isolate->factory()->NewStringFromTwoByte(Vector<const uint16_t>(
-          reinterpret_cast<const uint16_t*>(converted.getBuffer()),
-          converted.length())));
-  return *result;
+  typedef int32_t (*case_conversion_fn)(
+      UChar * dest, int32_t destCapacity, const UChar* src, int32_t srcLength,
+      const char* locale, UErrorCode* pErrorCode);
+  case_conversion_fn fn = is_to_upper ? u_strToUpper : u_strToLower;
+  const char* locale = locale_id == -1 ? "" : conversion_locales[locale_id];
+
+  int32_t dest_length = src_length;
+  UErrorCode error;
+  Handle<SeqTwoByteString> result;
+  do {
+    result =
+        isolate->factory()->NewRawTwoByteString(dest_length).ToHandleChecked();
+    base::SmartArrayPointer<uc16> sap;
+    DisallowHeapAllocation no_gc;
+    String::FlatContent flat = s->GetFlatContent();
+    // For OneByteString, |src| is already obtained with |sap| outside the loop.
+    if (flat.IsTwoByte())
+      src = reinterpret_cast<const UChar*>(flat.ToUC16Vector().start());
+    error = U_ZERO_ERROR;
+    dest_length = fn(reinterpret_cast<UChar*>(result->GetChars()), dest_length,
+                     src, src_length, locale, &error);
+  } while (error == U_BUFFER_OVERFLOW_ERROR);
+
+  // In most cases, the output will fill the destination buffer completely
+  // leading to an unterminated string (U_STRING_NOT_TERMINATED_WARNING).
+  // Only in rare cases, it'll be shorter than the destination buffer and
+  // |result| has to be truncated.
+  DCHECK(U_SUCCESS(error));
+  if (U_SUCCESS(error)) {
+    if (V8_UNLIKELY(error != U_STRING_NOT_TERMINATED_WARNING)) {
+      result = Handle<SeqTwoByteString>::cast(
+          SeqString::Truncate(result, dest_length));
+    }
+    return *result;
+  }
+  return *s;
 }
 
 inline bool IsASCIIUpper(uint16_t ch) { return ch >= 'A' && ch <= 'Z'; }