Fix state of reverse iterator after ZWJ

mbrubeck · mbrubeck · commit 3d593f8510f2 · 2017-02-06T10:06:39.000-08:00
Fixes unicode-rs#14.
diff --git a/src/grapheme.rs b/src/grapheme.rs
@@ -344,8 +344,8 @@ impl<'a> DoubleEndedIterator for Graphemes<'a> {
                     take_curr = false;
                     break;
                 },
-                Zwj => match cat {          // char to right is (GAZ|EBG)
-                    gr::GC_ZWJ => continue, // rule GB11: ZWJ x (GAZ|EBG)
+                Zwj => match cat {            // char to right is (GAZ|EBG)
+                    gr::GC_ZWJ => FindExtend, // rule GB11: ZWJ x (GAZ|EBG)
                     _ => {
                         take_curr = false;
                         break;
diff --git a/src/test.rs b/src/test.rs
@@ -30,7 +30,13 @@ fn test_graphemes() {
          &["\u{600}", "\u{20}", "\u{20}"]),
     ];
 
-    for &(s, g) in TEST_SAME.iter() {
+    pub const EXTRA_SAME: &'static [(&'static str, &'static [&'static str])] = &[
+        // family emoji (more than two emoji joined by ZWJ)
+        ("\u{1f468}\u{200d}\u{1f467}\u{200d}\u{1f466}",
+         &["\u{1f468}\u{200d}\u{1f467}\u{200d}\u{1f466}"]),
+    ];
+
+    for &(s, g) in TEST_SAME.iter().chain(EXTRA_SAME) {
         // test forward iterator
         assert!(UnicodeSegmentation::graphemes(s, true).eq(g.iter().cloned()));
         assert!(UnicodeSegmentation::graphemes(s, false).eq(g.iter().cloned()));
