Fix a bug in Prepend handling

mbrubeck · mbrubeck · commit 4acb5e9bb66d · 2016-12-27T10:17:05.000-08:00
The state was not updated correctly for the char after the Prepend.
diff --git a/Cargo.toml b/Cargo.toml
@@ -1,7 +1,7 @@
 [package]
 
 name = "unicode-segmentation"
-version = "1.0.0"
+version = "1.0.1"
 authors = ["kwantam <kwantam@gmail.com>"]
 
 homepage = "https://github.com/unicode-rs/unicode-segmentation"
diff --git a/src/grapheme.rs b/src/grapheme.rs
@@ -125,8 +125,11 @@ impl<'a> Iterator for Graphemes<'a> {
                     }
                     break;                      // rule GB4
                 }
-                Start => match cat {
-                    gr::GC_Control => break,
+                Start | Prepend => match cat {
+                    gr::GC_Control => {         // rule GB5
+                        take_curr = state == Start;
+                        break;
+                    }
                     gr::GC_L => HangulL,
                     gr::GC_LV | gr::GC_V => HangulLV,
                     gr::GC_LVT | gr::GC_T => HangulLVT,
@@ -163,13 +166,6 @@ impl<'a> Iterator for Graphemes<'a> {
                         break;
                     }
                 },
-                Prepend => match cat {      // rule GB9b
-                    gr::GC_Control => {
-                        take_curr = false;
-                        break;
-                    }
-                    _ => continue
-                },
                 Regional => match cat {     // rule GB12/GB13
                     gr::GC_Regional_Indicator => FindExtend,
                     _ => {
diff --git a/src/test.rs b/src/test.rs
@@ -23,6 +23,11 @@ fn test_graphemes() {
         ("\u{20}\u{600}\u{600}\u{20}",
          &["\u{20}", "\u{600}\u{600}\u{20}"],
          &["\u{20}", "\u{600}", "\u{600}", "\u{20}"]),
+
+	// Test for Prepend followed by two Any chars
+	("\u{600}\u{20}\u{20}",
+         &["\u{600}\u{20}", "\u{20}"],
+         &["\u{600}", "\u{20}", "\u{20}"]),
     ];
 
     for &(s, g) in TEST_SAME {
