Update CachingWordShapeIterator to more reliably handle emoji sequences

Instead of individual tests for tag sequences, use newer ICU methods to
detect xpicto_sequences as in Unicode TR#29, compare
https://www.unicode.org/reports/tr29/#Regex_Definitions by using ICU's
UCHAR_EMOJI_COMPONENT and UCHAR_EMOJI_COMPONENT checks.

Bug: 1121420
Change-Id: I37c9cab51ea88c6be85c2df637af03dcf68c568f
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2390341Reviewed-by: Koji Ishii <kojii@chromium.org>
Reviewed-by: Yoshifumi Inoue <yosin@chromium.org>
Commit-Queue: Dominik Röttsches <drott@chromium.org>
Cr-Commit-Position: refs/heads/master@{#804168}

third_party/blink/renderer/platform/fonts/shaping/caching_word_shape_iterator.h

@@ -125,16 +125,10 @@ class PLATFORM_EXPORT CachingWordShapeIterator final {
     bool has_any_script = !Character::IsCommonOrInheritedScript(ch);
     for (unsigned next_end = end; end < length; end = next_end) {
       ch = text_run_.CodepointAtAndNext(next_end);
-      // If ZWJ, include the next character.
-      if (ch == kZeroWidthJoinerCharacter) {
-        if (next_end < length)
-          text_run_.CodepointAtAndNext(next_end);
-        continue;
-      }
-      // Modifier check in order not to split those Emoji sequences.
+      // Modifier check in order not to split Emoji sequences.
       if (U_GET_GC_MASK(ch) & (U_GC_M_MASK | U_GC_LM_MASK | U_GC_SK_MASK) ||
-          Character::IsModifier(ch) || Character::IsEmojiTagSequence(ch) ||
-          ch == kCancelTag)
+          ch == kZeroWidthJoinerCharacter || Character::IsEmojiComponent(ch) ||
+          Character::IsExtendedPictographic(ch))
         continue;
       // Avoid delimiting COMMON/INHERITED alone, which makes harder to
       // identify the script.

third_party/blink/renderer/platform/fonts/shaping/caching_word_shaper_test.cc

@@ -213,82 +213,36 @@ TEST_F(CachingWordShaperTest, SegmentCJKAndNonCJKCommon) {
   ASSERT_FALSE(iterator.Next(&word_result));
 }
 
-TEST_F(CachingWordShaperTest, SegmentEmojiZWJCommon) {
-  // A family followed by a couple with heart emoji sequence,
-  // the latter including a variation selector.
-  const UChar kStr[] = {0xD83D, 0xDC68, 0x200D, 0xD83D, 0xDC69, 0x200D,
-                        0xD83D, 0xDC67, 0x200D, 0xD83D, 0xDC66, 0xD83D,
-                        0xDC69, 0x200D, 0x2764, 0xFE0F, 0x200D, 0xD83D,
-                        0xDC8B, 0x200D, 0xD83D, 0xDC68, 0x0};
-  TextRun text_run(kStr, 22);
-
-  scoped_refptr<const ShapeResult> word_result;
-  CachingWordShapeIterator iterator(cache.get(), text_run, &font);
-
-  ASSERT_TRUE(iterator.Next(&word_result));
-  EXPECT_EQ(22u, word_result->NumCharacters());
-
-  ASSERT_FALSE(iterator.Next(&word_result));
-}
-
-TEST_F(CachingWordShaperTest, SegmentEmojiZWJ) {
-  // ZWJ should include the next character in the "word", so  that they are
-  // shaped together.
-  String str(u"\U0001F3F4\u200D\u2620\uFE0F");
-  TextRun text_run(str);
-
-  scoped_refptr<const ShapeResult> word_result;
-  CachingWordShapeIterator iterator(cache.get(), text_run, &font);
-
-  ASSERT_TRUE(iterator.Next(&word_result));
-  EXPECT_EQ(str.length(), word_result->NumCharacters());
-
-  ASSERT_FALSE(iterator.Next(&word_result));
-}
-
-TEST_F(CachingWordShaperTest, SegmentEmojiPilotJudgeSequence) {
-  // A family followed by a couple with heart emoji sequence,
-  // the latter including a variation selector.
-  const UChar kStr[] = {0xD83D, 0xDC68, 0xD83C, 0xDFFB, 0x200D, 0x2696, 0xFE0F,
-                        0xD83D, 0xDC68, 0xD83C, 0xDFFB, 0x200D, 0x2708, 0xFE0F};
-  TextRun text_run(kStr, base::size(kStr));
-
-  scoped_refptr<const ShapeResult> word_result;
-  CachingWordShapeIterator iterator(cache.get(), text_run, &font);
-
-  ASSERT_TRUE(iterator.Next(&word_result));
-  EXPECT_EQ(base::size(kStr), word_result->NumCharacters());
-
-  ASSERT_FALSE(iterator.Next(&word_result));
-}
-
-TEST_F(CachingWordShaperTest, SegmentEmojiHeartZWJSequence) {
-  // A ZWJ, followed by two family ZWJ Sequences.
-  const UChar kStr[] = {0xD83D, 0xDC69, 0x200D, 0x2764, 0xFE0F, 0x200D,
-                        0xD83D, 0xDC8B, 0x200D, 0xD83D, 0xDC68, 0x0};
-  TextRun text_run(kStr, 11);
-
-  scoped_refptr<const ShapeResult> word_result;
-  CachingWordShapeIterator iterator(cache.get(), text_run, &font);
-
-  ASSERT_TRUE(iterator.Next(&word_result));
-  EXPECT_EQ(11u, word_result->NumCharacters());
-
-  ASSERT_FALSE(iterator.Next(&word_result));
-}
-
-TEST_F(CachingWordShaperTest, SegmentEmojiSignsOfHornsModifier) {
-  // A Sign of the Horns emoji, followed by a fitzpatrick modifer
-  const UChar kStr[] = {0xD83E, 0xDD18, 0xD83C, 0xDFFB, 0x0};
-  TextRun text_run(kStr, 4);
-
-  scoped_refptr<const ShapeResult> word_result;
-  CachingWordShapeIterator iterator(cache.get(), text_run, &font);
-
-  ASSERT_TRUE(iterator.Next(&word_result));
-  EXPECT_EQ(4u, word_result->NumCharacters());
-
-  ASSERT_FALSE(iterator.Next(&word_result));
+TEST_F(CachingWordShaperTest, SegmentEmojiSequences) {
+  std::vector<std::string> test_strings = {
+      // A family followed by a couple with heart emoji sequence,
+      // the latter including a variation selector.
+      u8"\U0001f468\u200D\U0001f469\u200D\U0001f467\u200D\U0001f466\U0001f469"
+      u8"\u200D\u2764\uFE0F\u200D\U0001f48b\u200D\U0001f468",
+      // Pirate flag
+      u8"\U0001F3F4\u200D\u2620\uFE0F",
+      // Pilot, judge sequence
+      u8"\U0001f468\U0001f3fb\u200D\u2696\uFE0F\U0001f468\U0001f3fb\u200D\u2708"
+      u8"\uFE0F",
+      // Woman, Kiss, Man sequence
+      u8"\U0001f469\u200D\u2764\uFE0F\u200D\U0001f48b\u200D\U0001f468",
+      // Signs of horns with skin tone modifier
+      u8"\U0001f918\U0001f3fb",
+      // Man, dark skin tone, red hair
+      u8"\U0001f468\U0001f3ff\u200D\U0001f9b0"};
+
+  for (auto test_string : test_strings) {
+    String emoji_string = String::FromUTF8(test_string);
+    TextRun text_run(emoji_string);
+    scoped_refptr<const ShapeResult> word_result;
+    CachingWordShapeIterator iterator(cache.get(), text_run, &font);
+
+    ASSERT_TRUE(iterator.Next(&word_result));
+    EXPECT_EQ(emoji_string.length(), word_result->NumCharacters())
+        << " Length mismatch for sequence: " << test_string;
+
+    ASSERT_FALSE(iterator.Next(&word_result));
+  }
 }
 
 TEST_F(CachingWordShaperTest, SegmentEmojiExtraZWJPrefix) {

third_party/blink/renderer/platform/text/character.cc

@@ -224,6 +224,14 @@ bool Character::IsEmojiTagSequence(UChar32 c) {
          (c >= kTagLatinSmallLetterA && c <= kTagLatinSmallLetterZ);
 }
 
+bool Character::IsExtendedPictographic(UChar32 c) {
+  return u_hasBinaryProperty(c, UCHAR_EXTENDED_PICTOGRAPHIC);
+}
+
+bool Character::IsEmojiComponent(UChar32 c) {
+  return u_hasBinaryProperty(c, UCHAR_EMOJI_COMPONENT);
+}
+
 template <typename CharacterType>
 static inline String NormalizeSpacesInternal(const CharacterType* characters,
                                              unsigned length) {

third_party/blink/renderer/platform/text/character.h

@@ -154,6 +154,8 @@ class PLATFORM_EXPORT Character {
   static bool IsModifier(UChar32 c) { return c >= 0x1F3FB && c <= 0x1F3FF; }
   // http://www.unicode.org/reports/tr51/proposed.html#flag-emoji-tag-sequences
   static bool IsEmojiTagSequence(UChar32);
+  static bool IsEmojiComponent(UChar32);
+  static bool IsExtendedPictographic(UChar32);
 
   static inline UChar NormalizeSpaces(UChar character) {
     if (TreatAsSpace(character))

third_party/blink/renderer/platform/text/character_test.cc

@@ -443,4 +443,28 @@ TEST(CharacterTest, IsVerticalMathCharacter) {
   }
 }
 
+TEST(CharacterTest, ExtendedPictographic) {
+  EXPECT_FALSE(Character::IsExtendedPictographic(0x00A8));
+  EXPECT_TRUE(Character::IsExtendedPictographic(0x00A9));
+  EXPECT_FALSE(Character::IsExtendedPictographic(0x00AA));
+  EXPECT_FALSE(Character::IsExtendedPictographic(0x3298));
+  EXPECT_TRUE(Character::IsExtendedPictographic(0x3299));
+  EXPECT_FALSE(Character::IsExtendedPictographic(0x329A));
+}
+
+TEST(CharacterTest, EmojiComponents) {
+  UChar32 false_set[] = {0x22,    0x2B,    0x29,    0x40,    0x200C,  0x200E,
+                         0x20E2,  0x20E4,  0xFE0E,  0xFE1A,  0x1F1E5, 0x1f200,
+                         0x1f3fa, 0x1f400, 0x1f9Af, 0x1f9b4, 0xe001F, 0xe0080};
+  UChar32 true_set[] = {0x23,    0x2a,    0x30,    0x39,    0x200d,
+                        0x20e3,  0xfe0f,  0x1f1e6, 0x1f1ff, 0x1f3fb,
+                        0x1f3ff, 0x1f9b0, 0x1f9b3, 0xe0020, 0xe007f};
+
+  for (auto false_test : false_set)
+    EXPECT_FALSE(Character::IsEmojiComponent(false_test));
+
+  for (auto true_test : true_set)
+    EXPECT_TRUE(Character::IsEmojiComponent(true_test));
+}
+
 }  // namespace blink