Revert "Re-enable Unicode ZWJ handling code that had been disabled pending an ICU update (#4035)" (#4036)

jason-simmons · web-flow · commit f9e00a7c726c · 2017-08-29T12:53:25.000-07:00
This reverts commit afd2e6b. Fuchsia is still using an older version of ICU that does not support this. Also temporarily disable a unit test that failed without the reverted patch.
diff --git a/third_party/txt/src/minikin/GraphemeBreak.cpp b/third_party/txt/src/minikin/GraphemeBreak.cpp
@@ -112,7 +112,8 @@ bool GraphemeBreak::isGraphemeBreak(const float* advances,
     return false;
   }
   // Rule GB9, x (Extend | ZWJ); Rule GB9a, x SpacingMark; Rule GB9b, Prepend x
-  if (p2 == U_GCB_EXTEND || p2 == U_GCB_ZWJ || p2 == U_GCB_SPACING_MARK ||
+  // TODO(abarth): Add U_GCB_ZWJ once we update ICU.
+  if (p2 == U_GCB_EXTEND || /* p2 == U_GCB_ZWJ || */ p2 == U_GCB_SPACING_MARK ||
       p1 == U_GCB_PREPEND) {
     return false;
   }
@@ -156,26 +157,25 @@ bool GraphemeBreak::isGraphemeBreak(const float* advances,
     }
   }
 
+  // TODO(abarth): Enablet his code once we update ICU.
   // Tailored version of Rule GB11, ZWJ × (Glue_After_Zwj | EBG)
   // We try to make emoji sequences with ZWJ a single grapheme cluster, but only
   // if they actually merge to one cluster. So we are more relaxed than the UAX
   // #29 rules in accepting any emoji character after the ZWJ, but are tighter
   // in that we only treat it as one cluster if a ligature is actually formed
-  // and we also require the character before the ZWJ to also be an emoji.
-  if (p1 == U_GCB_ZWJ && isEmoji(c2) && offset_back > start) {
-    // look at character before ZWJ to see that both can participate in an
-    // emoji zwj sequence
-    uint32_t c0 = 0;
-    size_t offset_backback = offset_back;
-    U16_PREV(buf, start, offset_backback, c0);
-    if (c0 == 0xFE0F && offset_backback > start) {
-      // skip over emoji variation selector
-      U16_PREV(buf, start, offset_backback, c0);
-    }
-    if (isEmoji(c0)) {
-      return false;
-    }
-  }
+  // and we also require the character before the ZWJ to also be an emoji. if
+  // (p1 == U_GCB_ZWJ && isEmoji(c2) && offset_back > start) {
+  //     // look at character before ZWJ to see that both can participate in an
+  //     emoji zwj sequence uint32_t c0 = 0; size_t offset_backback =
+  //     offset_back; U16_PREV(buf, start, offset_backback, c0); if (c0 ==
+  //     0xFE0F && offset_backback > start) {
+  //         // skip over emoji variation selector
+  //         U16_PREV(buf, start, offset_backback, c0);
+  //     }
+  //     if (isEmoji(c0)) {
+  //         return false;
+  //     }
+  // }
 
   // Tailored version of Rule GB12 and Rule GB13 that look at even-odd cases.
   // sot   (RI RI)*  RI x RI
diff --git a/third_party/txt/tests/GraphemeBreakTests.cpp b/third_party/txt/tests/GraphemeBreakTests.cpp
@@ -143,7 +143,9 @@ TEST(GraphemeBreak, rules) {
 
   // Rule GB9, x (Extend | ZWJ)
   EXPECT_FALSE(IsBreak("'a' | U+0301"));  // combining accent
-  EXPECT_FALSE(IsBreak("'a' | U+200D"));  // ZWJ
+  // TODO(jsimmons): re-enable this test when ICU has been updated in all
+  // Flutter platforms.
+  //EXPECT_FALSE(IsBreak("'a' | U+200D"));  // ZWJ
   // Rule GB9a, x SpacingMark
   EXPECT_FALSE(IsBreak("U+0915 | U+093E"));  // KA, AA (spacing mark)
   // Rule GB9b, Prepend x
