[Intl] Cutting 43K by removing Unibrow when ICU available

Making 43K of room for landing ICU64.

Size Change (on x64.release)
D8 before 23,683,192
D8 after 23,639,296
Reduce 43,896 bytes

Bugs: v8:8348

Change-Id: I057f7d59e955a2e5e017873e5b3b5daf5b142ae2
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1478710
Commit-Queue: Frank Tang <ftang@chromium.org>
Reviewed-by: Yang Guo <yangguo@chromium.org>
Cr-Commit-Position: refs/heads/master@{#60616}

src/isolate.h

@@ -970,6 +970,7 @@ class Isolate final : private HiddenFactory {
 
   ThreadManager* thread_manager() { return thread_manager_; }
 
+#ifndef V8_INTL_SUPPORT
   unibrow::Mapping<unibrow::Ecma262UnCanonicalize>* jsregexp_uncanonicalize() {
     return &jsregexp_uncanonicalize_;
   }
@@ -978,14 +979,15 @@ class Isolate final : private HiddenFactory {
     return &jsregexp_canonrange_;
   }
 
-  RuntimeState* runtime_state() { return &runtime_state_; }
-
-  Builtins* builtins() { return &builtins_; }
-
   unibrow::Mapping<unibrow::Ecma262Canonicalize>*
       regexp_macro_assembler_canonicalize() {
     return &regexp_macro_assembler_canonicalize_;
   }
+#endif  // !V8_INTL_SUPPORT
+
+  RuntimeState* runtime_state() { return &runtime_state_; }
+
+  Builtins* builtins() { return &builtins_; }
 
   RegExpStack* regexp_stack() { return regexp_stack_; }
 
@@ -996,11 +998,6 @@ class Isolate final : private HiddenFactory {
 
   std::vector<int>* regexp_indices() { return &regexp_indices_; }
 
-  unibrow::Mapping<unibrow::Ecma262Canonicalize>*
-      interp_canonicalize_mapping() {
-    return &regexp_macro_assembler_canonicalize_;
-  }
-
   Debug* debug() { return debug_; }
 
   bool* is_profiling_address() { return &is_profiling_; }
@@ -1642,10 +1639,12 @@ class Isolate final : private HiddenFactory {
   RuntimeState runtime_state_;
   Builtins builtins_;
   SetupIsolateDelegate* setup_delegate_ = nullptr;
+#ifndef V8_INTL_SUPPORT
   unibrow::Mapping<unibrow::Ecma262UnCanonicalize> jsregexp_uncanonicalize_;
   unibrow::Mapping<unibrow::CanonicalizationRange> jsregexp_canonrange_;
   unibrow::Mapping<unibrow::Ecma262Canonicalize>
       regexp_macro_assembler_canonicalize_;
+#endif  // !V8_INTL_SUPPORT
   RegExpStack* regexp_stack_ = nullptr;
   std::vector<int> regexp_indices_;
   DateCache* date_cache_ = nullptr;

src/regexp/interpreter-irregexp.cc

@@ -21,8 +21,6 @@
 namespace v8 {
 namespace internal {
 
-using Canonicalize = unibrow::Mapping<unibrow::Ecma262Canonicalize>;
-
 static bool BackRefMatchesNoCase(Isolate* isolate, int from, int current,
                                  int len, Vector<const uc16> subject,
                                  bool unicode) {

src/regexp/regexp-macro-assembler.cc

@@ -12,6 +12,7 @@
 
 #ifdef V8_INTL_SUPPORT
 #include "unicode/uchar.h"
+#include "unicode/unistr.h"
 #endif  // V8_INTL_SUPPORT
 
 namespace v8 {
@@ -33,37 +34,17 @@ int RegExpMacroAssembler::CaseInsensitiveCompareUC16(Address byte_offset1,
   // A GC might move the calling generated code and invalidate the
   // return address on the stack.
   DCHECK_EQ(0, byte_length % 2);
+
+#ifdef V8_INTL_SUPPORT
+  int32_t length = (int32_t)(byte_length >> 1);
+  icu::UnicodeString uni_str_1(reinterpret_cast<const char16_t*>(byte_offset1),
+                               length);
+  return uni_str_1.caseCompare(reinterpret_cast<const char16_t*>(byte_offset2),
+                               length, U_FOLD_CASE_DEFAULT) == 0;
+#else
   uc16* substring1 = reinterpret_cast<uc16*>(byte_offset1);
   uc16* substring2 = reinterpret_cast<uc16*>(byte_offset2);
   size_t length = byte_length >> 1;
-
-#ifdef V8_INTL_SUPPORT
-  if (isolate == nullptr) {
-    for (size_t i = 0; i < length; i++) {
-      uc32 c1 = substring1[i];
-      uc32 c2 = substring2[i];
-      if (unibrow::Utf16::IsLeadSurrogate(c1)) {
-        // Non-BMP characters do not have case-equivalents in the BMP.
-        // Both have to be non-BMP for them to be able to match.
-        if (!unibrow::Utf16::IsLeadSurrogate(c2)) return 0;
-        if (i + 1 < length) {
-          uc16 c1t = substring1[i + 1];
-          uc16 c2t = substring2[i + 1];
-          if (unibrow::Utf16::IsTrailSurrogate(c1t) &&
-              unibrow::Utf16::IsTrailSurrogate(c2t)) {
-            c1 = unibrow::Utf16::CombineSurrogatePair(c1, c1t);
-            c2 = unibrow::Utf16::CombineSurrogatePair(c2, c2t);
-            i++;
-          }
-        }
-      }
-      c1 = u_foldCase(c1, U_FOLD_CASE_DEFAULT);
-      c2 = u_foldCase(c2, U_FOLD_CASE_DEFAULT);
-      if (c1 != c2) return 0;
-    }
-    return 1;
-  }
-#endif  // V8_INTL_SUPPORT
   DCHECK_NOT_NULL(isolate);
   unibrow::Mapping<unibrow::Ecma262Canonicalize>* canonicalize =
       isolate->regexp_macro_assembler_canonicalize();
@@ -83,6 +64,7 @@ int RegExpMacroAssembler::CaseInsensitiveCompareUC16(Address byte_offset1,
     }
   }
   return 1;
+#endif  // V8_INTL_SUPPORT
 }
 
 

src/unicode-inl.h

@@ -11,6 +11,7 @@
 
 namespace unibrow {
 
+#ifndef V8_INTL_SUPPORT
 template <class T, int s> bool Predicate<T, s>::get(uchar code_point) {
   CacheEntry entry = entries_[code_point & kMask];
   if (entry.code_point() == code_point) return entry.value();
@@ -55,6 +56,7 @@ template <class T, int s> int Mapping<T, s>::CalculateValue(uchar c, uchar n,
     return length;
   }
 }
+#endif  // !V8_INTL_SUPPORT
 
 // Decodes UTF-8 bytes incrementally, allowing the decoding of bytes as they
 // stream in. This **must** be followed by a call to ValueOfIncrementalFinish

src/unicode.cc

@@ -15,8 +15,11 @@
 
 namespace unibrow {
 
+#ifndef V8_INTL_SUPPORT
 static const int kStartBit = (1 << 30);
 static const int kChunkBits = (1 << 13);
+#endif  // !V8_INTL_SUPPORT
+
 static const uchar kSentinel = static_cast<uchar>(-1);
 
 /**
@@ -28,7 +31,7 @@ typedef signed short int16_t;  // NOLINT
 typedef unsigned short uint16_t;  // NOLINT
 typedef int int32_t;  // NOLINT
 
-
+#ifndef V8_INTL_SUPPORT
 // All access to the character table should go through this function.
 template <int D>
 static inline uchar TableGet(const int32_t* table, int index) {
@@ -44,7 +47,6 @@ static inline bool IsStart(int32_t entry) {
   return (entry & kStartBit) != 0;
 }
 
-#ifndef V8_INTL_SUPPORT
 /**
  * Look up a character in the Unicode table using a mix of binary and
  * interpolation search.  For a uniformly distributed array
@@ -92,6 +94,7 @@ struct MultiCharacterSpecialCase {
   uchar chars[kW];
 };
 
+#ifndef V8_INTL_SUPPORT
 // Look up the mapping for the given character in the specified table,
 // which is of the specified length and uses the specified special case
 // mapping for multi-char mappings.  The next parameter is the character
@@ -192,6 +195,7 @@ static int LookupMapping(const int32_t* table,
     return 0;
   }
 }
+#endif  // !V8_INTL_SUPPORT
 
 // This method decodes an UTF-8 value according to RFC 3629 and
 // https://encoding.spec.whatwg.org/#utf-8-decoder .
@@ -1596,7 +1600,6 @@ int ToUppercase::Convert(uchar c,
     default: return 0;
   }
 }
-#endif  // !V8_INTL_SUPPORT
 
 static const MultiCharacterSpecialCase<1> kEcma262CanonicalizeMultiStrings0[1] = {  // NOLINT
   {{kSentinel}} }; // NOLINT
@@ -3072,98 +3075,75 @@ int CanonicalizationRange::Convert(uchar c,
   }
 }
 
-
 const uchar UnicodeData::kMaxCodePoint = 0xFFFD;
 
 int UnicodeData::GetByteCount() {
-#ifndef V8_INTL_SUPPORT                                 // NOLINT
-  return kUppercaseTable0Size * sizeof(int32_t)         // NOLINT
-         + kUppercaseTable1Size * sizeof(int32_t)       // NOLINT
-         + kUppercaseTable5Size * sizeof(int32_t)       // NOLINT
-         + kUppercaseTable7Size * sizeof(int32_t)       // NOLINT
-         + kLetterTable0Size * sizeof(int32_t)          // NOLINT
-         + kLetterTable1Size * sizeof(int32_t)          // NOLINT
-         + kLetterTable2Size * sizeof(int32_t)          // NOLINT
-         + kLetterTable3Size * sizeof(int32_t)          // NOLINT
-         + kLetterTable4Size * sizeof(int32_t)          // NOLINT
-         + kLetterTable5Size * sizeof(int32_t)          // NOLINT
-         + kLetterTable6Size * sizeof(int32_t)          // NOLINT
-         + kLetterTable7Size * sizeof(int32_t)          // NOLINT
-         + kID_StartTable0Size * sizeof(int32_t)        // NOLINT
-         + kID_StartTable1Size * sizeof(int32_t)        // NOLINT
-         + kID_StartTable2Size * sizeof(int32_t)        // NOLINT
-         + kID_StartTable3Size * sizeof(int32_t)        // NOLINT
-         + kID_StartTable4Size * sizeof(int32_t)        // NOLINT
-         + kID_StartTable5Size * sizeof(int32_t)        // NOLINT
-         + kID_StartTable6Size * sizeof(int32_t)        // NOLINT
-         + kID_StartTable7Size * sizeof(int32_t)        // NOLINT
-         + kID_ContinueTable0Size * sizeof(int32_t)     // NOLINT
-         + kID_ContinueTable1Size * sizeof(int32_t)     // NOLINT
-         + kID_ContinueTable5Size * sizeof(int32_t)     // NOLINT
-         + kID_ContinueTable7Size * sizeof(int32_t)     // NOLINT
-         + kWhiteSpaceTable0Size * sizeof(int32_t)      // NOLINT
-         + kWhiteSpaceTable1Size * sizeof(int32_t)      // NOLINT
-         + kWhiteSpaceTable7Size * sizeof(int32_t)      // NOLINT
-         +
-         kToLowercaseMultiStrings0Size *
-             sizeof(MultiCharacterSpecialCase<2>)  // NOLINT
-         +
-         kToLowercaseMultiStrings1Size *
-             sizeof(MultiCharacterSpecialCase<1>)  // NOLINT
-         +
-         kToLowercaseMultiStrings5Size *
-             sizeof(MultiCharacterSpecialCase<1>)  // NOLINT
-         +
-         kToLowercaseMultiStrings7Size *
-             sizeof(MultiCharacterSpecialCase<1>)  // NOLINT
-         +
-         kToUppercaseMultiStrings0Size *
-             sizeof(MultiCharacterSpecialCase<3>)  // NOLINT
-         +
-         kToUppercaseMultiStrings1Size *
-             sizeof(MultiCharacterSpecialCase<1>)  // NOLINT
-         +
-         kToUppercaseMultiStrings5Size *
-             sizeof(MultiCharacterSpecialCase<1>)  // NOLINT
-         +
-         kToUppercaseMultiStrings7Size *
-             sizeof(MultiCharacterSpecialCase<3>)  // NOLINT
-#else
-  return
-#endif  // !V8_INTL_SUPPORT
-         +
-         kEcma262CanonicalizeMultiStrings0Size *
-             sizeof(MultiCharacterSpecialCase<1>)  // NOLINT
-         +
-         kEcma262CanonicalizeMultiStrings1Size *
-             sizeof(MultiCharacterSpecialCase<1>)  // NOLINT
-         +
-         kEcma262CanonicalizeMultiStrings5Size *
-             sizeof(MultiCharacterSpecialCase<1>)  // NOLINT
-         +
-         kEcma262CanonicalizeMultiStrings7Size *
-             sizeof(MultiCharacterSpecialCase<1>)  // NOLINT
-         +
-         kEcma262UnCanonicalizeMultiStrings0Size *
-             sizeof(MultiCharacterSpecialCase<4>)  // NOLINT
-         +
-         kEcma262UnCanonicalizeMultiStrings1Size *
-             sizeof(MultiCharacterSpecialCase<2>)  // NOLINT
-         +
-         kEcma262UnCanonicalizeMultiStrings5Size *
-             sizeof(MultiCharacterSpecialCase<2>)  // NOLINT
-         +
-         kEcma262UnCanonicalizeMultiStrings7Size *
-             sizeof(MultiCharacterSpecialCase<2>)  // NOLINT
-         +
-         kCanonicalizationRangeMultiStrings0Size *
-             sizeof(MultiCharacterSpecialCase<1>)  // NOLINT
-         +
-         kCanonicalizationRangeMultiStrings1Size *
-             sizeof(MultiCharacterSpecialCase<1>)  // NOLINT
-         +
-         kCanonicalizationRangeMultiStrings7Size *
-             sizeof(MultiCharacterSpecialCase<1>);  // NOLINT
+  return kUppercaseTable0Size * sizeof(int32_t)      // NOLINT
+         + kUppercaseTable1Size * sizeof(int32_t)    // NOLINT
+         + kUppercaseTable5Size * sizeof(int32_t)    // NOLINT
+         + kUppercaseTable7Size * sizeof(int32_t)    // NOLINT
+         + kLetterTable0Size * sizeof(int32_t)       // NOLINT
+         + kLetterTable1Size * sizeof(int32_t)       // NOLINT
+         + kLetterTable2Size * sizeof(int32_t)       // NOLINT
+         + kLetterTable3Size * sizeof(int32_t)       // NOLINT
+         + kLetterTable4Size * sizeof(int32_t)       // NOLINT
+         + kLetterTable5Size * sizeof(int32_t)       // NOLINT
+         + kLetterTable6Size * sizeof(int32_t)       // NOLINT
+         + kLetterTable7Size * sizeof(int32_t)       // NOLINT
+         + kID_StartTable0Size * sizeof(int32_t)     // NOLINT
+         + kID_StartTable1Size * sizeof(int32_t)     // NOLINT
+         + kID_StartTable2Size * sizeof(int32_t)     // NOLINT
+         + kID_StartTable3Size * sizeof(int32_t)     // NOLINT
+         + kID_StartTable4Size * sizeof(int32_t)     // NOLINT
+         + kID_StartTable5Size * sizeof(int32_t)     // NOLINT
+         + kID_StartTable6Size * sizeof(int32_t)     // NOLINT
+         + kID_StartTable7Size * sizeof(int32_t)     // NOLINT
+         + kID_ContinueTable0Size * sizeof(int32_t)  // NOLINT
+         + kID_ContinueTable1Size * sizeof(int32_t)  // NOLINT
+         + kID_ContinueTable5Size * sizeof(int32_t)  // NOLINT
+         + kID_ContinueTable7Size * sizeof(int32_t)  // NOLINT
+         + kWhiteSpaceTable0Size * sizeof(int32_t)   // NOLINT
+         + kWhiteSpaceTable1Size * sizeof(int32_t)   // NOLINT
+         + kWhiteSpaceTable7Size * sizeof(int32_t)   // NOLINT
+         + kToLowercaseMultiStrings0Size *
+               sizeof(MultiCharacterSpecialCase<2>)  // NOLINT
+         + kToLowercaseMultiStrings1Size *
+               sizeof(MultiCharacterSpecialCase<1>)  // NOLINT
+         + kToLowercaseMultiStrings5Size *
+               sizeof(MultiCharacterSpecialCase<1>)  // NOLINT
+         + kToLowercaseMultiStrings7Size *
+               sizeof(MultiCharacterSpecialCase<1>)  // NOLINT
+         + kToUppercaseMultiStrings0Size *
+               sizeof(MultiCharacterSpecialCase<3>)  // NOLINT
+         + kToUppercaseMultiStrings1Size *
+               sizeof(MultiCharacterSpecialCase<1>)  // NOLINT
+         + kToUppercaseMultiStrings5Size *
+               sizeof(MultiCharacterSpecialCase<1>)  // NOLINT
+         + kToUppercaseMultiStrings7Size *
+               sizeof(MultiCharacterSpecialCase<3>)  // NOLINT
+         + kEcma262CanonicalizeMultiStrings0Size *
+               sizeof(MultiCharacterSpecialCase<1>)  // NOLINT
+         + kEcma262CanonicalizeMultiStrings1Size *
+               sizeof(MultiCharacterSpecialCase<1>)  // NOLINT
+         + kEcma262CanonicalizeMultiStrings5Size *
+               sizeof(MultiCharacterSpecialCase<1>)  // NOLINT
+         + kEcma262CanonicalizeMultiStrings7Size *
+               sizeof(MultiCharacterSpecialCase<1>)  // NOLINT
+         + kEcma262UnCanonicalizeMultiStrings0Size *
+               sizeof(MultiCharacterSpecialCase<4>)  // NOLINT
+         + kEcma262UnCanonicalizeMultiStrings1Size *
+               sizeof(MultiCharacterSpecialCase<2>)  // NOLINT
+         + kEcma262UnCanonicalizeMultiStrings5Size *
+               sizeof(MultiCharacterSpecialCase<2>)  // NOLINT
+         + kEcma262UnCanonicalizeMultiStrings7Size *
+               sizeof(MultiCharacterSpecialCase<2>)  // NOLINT
+         + kCanonicalizationRangeMultiStrings0Size *
+               sizeof(MultiCharacterSpecialCase<1>)  // NOLINT
+         + kCanonicalizationRangeMultiStrings1Size *
+               sizeof(MultiCharacterSpecialCase<1>)  // NOLINT
+         + kCanonicalizationRangeMultiStrings7Size *
+               sizeof(MultiCharacterSpecialCase<1>);  // NOLINT
 }
+#endif  // !V8_INTL_SUPPORT
 
 }  // namespace unibrow

src/unicode.h

@@ -25,6 +25,7 @@ typedef unsigned char byte;
  */
 const int kMaxMappingSize = 4;
 
+#ifndef V8_INTL_SUPPORT
 template <class T, int size = 256>
 class Predicate {
  public:
@@ -87,7 +88,6 @@ class Mapping {
   CacheEntry entries_[kSize];
 };
 
-
 class UnicodeData {
  private:
   friend class Test;
@@ -95,6 +95,7 @@ class UnicodeData {
   static const uchar kMaxCodePoint;
 };
 
+#endif  // !V8_INTL_SUPPORT
 
 class Utf16 {
  public:
@@ -227,7 +228,6 @@ struct ToUppercase {
                      uchar* result,
                      bool* allow_caching_ptr);
 };
-#endif
 struct Ecma262Canonicalize {
   static const int kMaxWidth = 1;
   static int Convert(uchar c,
@@ -249,6 +249,7 @@ struct CanonicalizationRange {
                      uchar* result,
                      bool* allow_caching_ptr);
 };
+#endif  // !V8_INTL_SUPPORT
 
 }  // namespace unibrow
 

test/cctest/test-regexp.cc

@@ -1488,7 +1488,7 @@ TEST(AddInverseToTable) {
   CHECK(table.Get(0xFFFF)->Get(0));
 }
 
-
+#ifndef V8_INTL_SUPPORT
 static uc32 canonicalize(uc32 c) {
   unibrow::uchar canon[unibrow::Ecma262Canonicalize::kMaxWidth];
   int count = unibrow::Ecma262Canonicalize::Convert(c, '\0', canon, nullptr);
@@ -1500,7 +1500,6 @@ static uc32 canonicalize(uc32 c) {
   }
 }
 
-
 TEST(LatinCanonicalize) {
   unibrow::Mapping<unibrow::Ecma262UnCanonicalize> un_canonicalize;
   for (unibrow::uchar lower = 'a'; lower <= 'z'; lower++) {
@@ -1514,7 +1513,6 @@ TEST(LatinCanonicalize) {
   }
   for (uc32 c = 128; c < (1 << 21); c++)
     CHECK_GE(canonicalize(c), 128);
-#ifndef V8_INTL_SUPPORT
   unibrow::Mapping<unibrow::ToUppercase> to_upper;
   // Canonicalization is only defined for the Basic Multilingual Plane.
   for (uc32 c = 0; c < (1 << 16); c++) {
@@ -1529,10 +1527,8 @@ TEST(LatinCanonicalize) {
       u = c;
     CHECK_EQ(u, canonicalize(c));
   }
-#endif
 }
 
-
 static uc32 CanonRangeEnd(uc32 c) {
   unibrow::uchar canon[unibrow::CanonicalizationRange::kMaxWidth];
   int count = unibrow::CanonicalizationRange::Convert(c, '\0', canon, nullptr);
@@ -1588,6 +1584,7 @@ TEST(UncanonicalizeEquivalence) {
   }
 }
 
+#endif
 
 static void TestRangeCaseIndependence(Isolate* isolate, CharacterRange input,
                                       Vector<CharacterRange> expected) {
@@ -1621,21 +1618,26 @@ TEST(CharacterRangeCaseIndependence) {
                                   CharacterRange::Singleton('A'));
   TestSimpleRangeCaseIndependence(isolate, CharacterRange::Singleton('z'),
                                   CharacterRange::Singleton('Z'));
+#ifndef V8_INTL_SUPPORT
   TestSimpleRangeCaseIndependence(isolate, CharacterRange::Range('a', 'z'),
                                   CharacterRange::Range('A', 'Z'));
+#endif  // !V8_INTL_SUPPORT
   TestSimpleRangeCaseIndependence(isolate, CharacterRange::Range('c', 'f'),
                                   CharacterRange::Range('C', 'F'));
   TestSimpleRangeCaseIndependence(isolate, CharacterRange::Range('a', 'b'),
                                   CharacterRange::Range('A', 'B'));
   TestSimpleRangeCaseIndependence(isolate, CharacterRange::Range('y', 'z'),
                                   CharacterRange::Range('Y', 'Z'));
+#ifndef V8_INTL_SUPPORT
   TestSimpleRangeCaseIndependence(isolate,
                                   CharacterRange::Range('a' - 1, 'z' + 1),
                                   CharacterRange::Range('A', 'Z'));
   TestSimpleRangeCaseIndependence(isolate, CharacterRange::Range('A', 'Z'),
                                   CharacterRange::Range('a', 'z'));
+#endif  // !V8_INTL_SUPPORT
   TestSimpleRangeCaseIndependence(isolate, CharacterRange::Range('C', 'F'),
                                   CharacterRange::Range('c', 'f'));
+#ifndef V8_INTL_SUPPORT
   TestSimpleRangeCaseIndependence(isolate,
                                   CharacterRange::Range('A' - 1, 'Z' + 1),
                                   CharacterRange::Range('a', 'z'));
@@ -1644,6 +1646,7 @@ TEST(CharacterRangeCaseIndependence) {
   // whole block at a time.
   TestSimpleRangeCaseIndependence(isolate, CharacterRange::Range('A', 'k'),
                                   CharacterRange::Range('a', 'z'));
+#endif  // !V8_INTL_SUPPORT
 }
 
 

test/intl/regress-8348.js

+// Copyright 2019 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.
+
+function testRegExpI(text, msg) {
+  assertTrue(new RegExp(text, 'i').test(text.toUpperCase()), msg + ': ' + text);
+}
+
+testRegExpI('abc', 'ASCII');
+testRegExpI('ABC', 'ASCII');
+testRegExpI('rst', 'ASCII');
+testRegExpI('RST', 'ASCII');
+
+testRegExpI('αβψδεφ', 'Greek');
+
+testRegExpI('\u1c80\u1c81', 'Historic Cyrillic added in Unicode 9');
+testRegExpI('\u026A', 'Dotless I, uppercase form added in Unicode 9');
+
+testRegExpI('ოქტ', 'Georgian Mtavruli added in Unicode 11');