Update Unicode data in Character class

This patch updates Unicode data in Character class:
1. UAX#50 Unicode Vertical Text[1] to rev 19 (Unicode 10).
2. IsCJKIdeographOrSymbol was not updated for a while, include
   ideographic characters to be in sync with UAX#50.
3. IsCJKIdeographOrSymbol includes all plane 2 rather than each
   block.

[1] http://www.unicode.org/reports/tr50/

Change-Id: Ie8e7f3b45d340870be5d1316021b49231be9cdd2
Reviewed-on: https://chromium-review.googlesource.com/822551
Commit-Queue: Koji Ishii <kojii@chromium.org>
Reviewed-by: Dominik Röttsches <drott@chromium.org>
Reviewed-by: Jungshik Shin <jshin@chromium.org>
Cr-Commit-Position: refs/heads/master@{#524661}

third_party/WebKit/Source/platform/text/Character.cpp

@@ -110,11 +110,7 @@ bool Character::IsUprightInMixedVertical(UChar32 character) {
   RETURN_HAS_PROPERTY(character, kIsUprightInMixedVertical)
 }
 
-bool Character::IsCJKIdeographOrSymbol(UChar32 c) {
-  // Likely common case
-  if (c < 0x2C7)
-    return false;
-
+bool Character::IsCJKIdeographOrSymbolSlow(UChar32 c) {
   RETURN_HAS_PROPERTY(c, kIsCJKIdeographOrSymbol)
 }
 
@@ -219,16 +215,12 @@ bool Character::CanTextDecorationSkipInk(UChar32 codepoint) {
   switch (block) {
     // These blocks contain CJK characters we don't want to skip ink, but are
     // not ideograph that IsCJKIdeographOrSymbol() does not cover.
-    case UBLOCK_BOPOMOFO:
-    case UBLOCK_BOPOMOFO_EXTENDED:
     case UBLOCK_HANGUL_JAMO:
     case UBLOCK_HANGUL_COMPATIBILITY_JAMO:
     case UBLOCK_HANGUL_SYLLABLES:
     case UBLOCK_HANGUL_JAMO_EXTENDED_A:
     case UBLOCK_HANGUL_JAMO_EXTENDED_B:
     case UBLOCK_LINEAR_B_IDEOGRAMS:
-    case UBLOCK_TANGUT:
-    case UBLOCK_TANGUT_COMPONENTS:
       return false;
     default:
       return true;

third_party/WebKit/Source/platform/text/Character.h

@@ -63,7 +63,10 @@ class PLATFORM_EXPORT Character {
                         0xE01EF);  // VARIATION SELECTOR-17 to 256
   }
 
-  static bool IsCJKIdeographOrSymbol(UChar32);
+  static bool IsCJKIdeographOrSymbol(UChar32 c) {
+    // Below U+02C7 is likely a common case.
+    return c < 0x2C7 ? false : IsCJKIdeographOrSymbolSlow(c);
+  }
   static bool IsCJKIdeographOrSymbolBase(UChar32 c) {
     return IsCJKIdeographOrSymbol(c) &&
            !(U_GET_GC_MASK(c) & (U_GC_M_MASK | U_GC_LM_MASK | U_GC_SK_MASK));
@@ -172,6 +175,9 @@ class PLATFORM_EXPORT Character {
   static String NormalizeSpaces(const UChar*, unsigned length);
 
   static bool IsCommonOrInheritedScript(UChar32);
+
+ private:
+  static bool IsCJKIdeographOrSymbolSlow(UChar32);
 };
 
 }  // namespace blink

third_party/WebKit/Source/platform/text/CharacterPropertyDataGenerator.h

@@ -47,13 +47,10 @@ static const UChar32 kIsCJKIdeographOrSymbolRanges[] = {
     0x4E00, 0x9FFF,
     // CJK Compatibility Ideographs.
     0xF900, 0xFAFF,
-    // CJK Unified Ideographs Extension B.
-    0x20000, 0x2A6DF,
-    // CJK Unified Ideographs Extension C.
-    // CJK Unified Ideographs Extension D.
-    0x2A700, 0x2B81F,
+    // Unicode Plane 2: Supplementary Ideographic Plane. This plane includes:
+    // CJK Unified Ideographs Extension B to F.
     // CJK Compatibility Ideographs Supplement.
-    0x2F800, 0x2FA1F,
+    0x20000, 0x2FFFF,
 
     // cjkSymbolRanges
     0x2156, 0x215A, 0x2160, 0x216B, 0x2170, 0x217B, 0x231A, 0x231B, 0x23E9,
@@ -80,13 +77,29 @@ static const UChar32 kIsCJKIdeographOrSymbolRanges[] = {
     0x3190, 0x31BF,
     // Enclosed CJK Letters and Months (0x3200 .. 0x32FF).
     // CJK Compatibility (0x3300 .. 0x33FF).
-    0x3200, 0x33FF, 0xF860, 0xF862,
+    0x3200, 0x33FF,
+    // Yijing Hexagram Symbols
+    0x4DC0, 0x4DFF,
+    // http://www.unicode.org/Public/MAPPINGS/VENDORS/APPLE/JAPANESE.TXT
+    0xF860, 0xF862,
     // CJK Compatibility Forms.
     // Small Form Variants (for CNS 11643).
     0xFE30, 0xFE6F,
     // Halfwidth and Fullwidth Forms
     // Usually only used in CJK
     0xFF00, 0xFF0C, 0xFF0E, 0xFF1A, 0xFF1F, 0xFFEF,
+    // Ideographic Symbols and Punctuation
+    0x16FE0, 0x16FFF,
+    // Tangut
+    0x17000, 0x187FF,
+    // Tangut Components
+    0x18800, 0x18AFF,
+    // Kana Supplement
+    0x1B000, 0x1B0FF,
+    // Kana Extended-A
+    0x1B100, 0x1B12F,
+    // Nushu
+    0x1B170, 0x1B2FF,
     // Emoji.
     0x1F110, 0x1F129, 0x1F130, 0x1F149, 0x1F150, 0x1F169, 0x1F170, 0x1F189,
     0x1F191, 0x1F19A, 0x1F1E6, 0x1F1FF, 0x1F200, 0x1F6FF,
@@ -158,6 +171,8 @@ static const UChar32 kIsUprightInMixedVerticalRanges[] = {
     0x10980, 0x1099F,
     // Siddham
     0x11580, 0x115FF,
+    // Zanabazar Square
+    0x11A00, 0x11AAF,
     // Egyptian Hieroglyphs
     0x13000, 0x1342F,
     // Anatolian Hieroglyphs
@@ -170,6 +185,10 @@ static const UChar32 kIsUprightInMixedVerticalRanges[] = {
     0x18800, 0x18AFF,
     // Kana Supplement
     0x1B000, 0x1B0FF,
+    // Kana Extended-A
+    0x1B100, 0x1B12F,
+    // Nushu
+    0x1B170, 0x1B2FF,
     // Byzantine Musical Symbols/Musical Symbols
     0x1D000, 0x1D1FF,
     // Tai Xuan Jing Symbols/Counting Rod Numerals

third_party/WebKit/Source/platform/text/CharacterTest.cpp

@@ -34,21 +34,30 @@ TEST(CharacterTest, HammerEmojiVsCJKIdeographOrSymbol) {
 
 static void TestSpecificUChar32RangeIdeograph(UChar32 range_start,
                                               UChar32 range_end,
-                                              bool before = true) {
-  if (before)
-    EXPECT_FALSE(Character::IsCJKIdeographOrSymbol(range_start - 1));
-  EXPECT_TRUE(Character::IsCJKIdeographOrSymbol(range_start));
-  EXPECT_TRUE(Character::IsCJKIdeographOrSymbol(
-      (UChar32)((uint64_t)range_start + (uint64_t)range_end) / 2));
-  EXPECT_TRUE(Character::IsCJKIdeographOrSymbol(range_end));
-  EXPECT_FALSE(Character::IsCJKIdeographOrSymbol(range_end + 1));
+                                              bool before = true,
+                                              bool after = true) {
+  if (before) {
+    EXPECT_FALSE(Character::IsCJKIdeographOrSymbol(range_start - 1))
+        << std::hex << (range_start - 1);
+  }
+  EXPECT_TRUE(Character::IsCJKIdeographOrSymbol(range_start))
+      << std::hex << range_start;
+  UChar32 mid = static_cast<UChar32>(
+      (static_cast<uint64_t>(range_start) + range_end) / 2);
+  EXPECT_TRUE(Character::IsCJKIdeographOrSymbol(mid)) << std::hex << mid;
+  EXPECT_TRUE(Character::IsCJKIdeographOrSymbol(range_end))
+      << std::hex << range_end;
+  if (after) {
+    EXPECT_FALSE(Character::IsCJKIdeographOrSymbol(range_end + 1))
+        << std::hex << (range_end + 1);
+  }
 }
 
 TEST(CharacterTest, TestIsCJKIdeograph) {
   // The basic CJK Unified Ideographs block.
-  TestSpecificUChar32RangeIdeograph(0x4E00, 0x9FFF);
+  TestSpecificUChar32RangeIdeograph(0x4E00, 0x9FFF, false);
   // CJK Unified Ideographs Extension A.
-  TestSpecificUChar32RangeIdeograph(0x3400, 0x4DBF, false);
+  TestSpecificUChar32RangeIdeograph(0x3400, 0x4DBF, false, false);
   // CJK Unified Ideographs Extension A and Kangxi Radicals.
   TestSpecificUChar32RangeIdeograph(0x2E80, 0x2FDF);
   // CJK Strokes.
@@ -56,12 +65,12 @@ TEST(CharacterTest, TestIsCJKIdeograph) {
   // CJK Compatibility Ideographs.
   TestSpecificUChar32RangeIdeograph(0xF900, 0xFAFF);
   // CJK Unified Ideographs Extension B.
-  TestSpecificUChar32RangeIdeograph(0x20000, 0x2A6DF);
+  TestSpecificUChar32RangeIdeograph(0x20000, 0x2A6DF, true, false);
   // CJK Unified Ideographs Extension C.
   // CJK Unified Ideographs Extension D.
-  TestSpecificUChar32RangeIdeograph(0x2A700, 0x2B81F);
+  TestSpecificUChar32RangeIdeograph(0x2A700, 0x2B81F, false, false);
   // CJK Compatibility Ideographs Supplement.
-  TestSpecificUChar32RangeIdeograph(0x2F800, 0x2FA1F);
+  TestSpecificUChar32RangeIdeograph(0x2F800, 0x2FA1F, false, false);
 }
 
 static void TestSpecificUChar32RangeIdeographSymbol(UChar32 range_start,
@@ -218,6 +227,23 @@ TEST(CharacterTest, TestIsCJKIdeographOrSymbol) {
   TestSpecificUChar32RangeIdeographSymbol(0x1F1E6, 0x1F6FF);
 }
 
+TEST(CharacterTest, CanTextDecorationSkipInk) {
+  // ASCII
+  EXPECT_TRUE(Character::CanTextDecorationSkipInk('a'));
+  // Hangul Jamo
+  EXPECT_FALSE(Character::CanTextDecorationSkipInk(0x1100));
+  // Hiragana
+  EXPECT_FALSE(Character::CanTextDecorationSkipInk(0x3041));
+  // Bopomofo
+  EXPECT_FALSE(Character::CanTextDecorationSkipInk(0x31A0));
+  // The basic CJK Unified Ideographs block
+  EXPECT_FALSE(Character::CanTextDecorationSkipInk(0x4E01));
+  // Hangul Syllables
+  EXPECT_FALSE(Character::CanTextDecorationSkipInk(0xAC00));
+  // Plane 2 / CJK Ideograph Extension B
+  EXPECT_FALSE(Character::CanTextDecorationSkipInk(0x20000));
+}
+
 TEST(CharacterTest, TestEmojiTextDefault) {
   // Text-default emoji, i.e.
   // Emoji=Yes and EmojiPresentation=No