Cache prepend lookbehinds

Author: Manishearth

src/grapheme.rs

@@ -223,6 +223,9 @@ impl<'a> DoubleEndedIterator for Graphemes<'a> {
         let mut state = Start;
         let mut cat = gr::GC_Any;
 
+        // caching used by next() should be invalidated
+        self.cat = None;
+
         'outer: for (curr, ch) in self.string.char_indices().rev() {
             previdx = idx;
             idx = curr;
@@ -365,10 +368,12 @@ impl<'a> DoubleEndedIterator for Graphemes<'a> {
         if self.extended && cat != gr::GC_Control {
             // rule GB9b: include any preceding Prepend characters
             for (i, c) in self.string[..idx].char_indices().rev() {
-                // TODO: Cache this to avoid repeated lookups in the common case.
                 match gr::grapheme_category(c) {
                     gr::GC_Prepend => idx = i,
-                    _ => break
+                    cat => {
+                        self.catb = Some(cat);
+                        break;
+                    }
                 }
             }
         }

src/test.rs

@@ -16,6 +16,15 @@ use std::prelude::v1::*;
 fn test_graphemes() {
     use testdata::{TEST_SAME, TEST_DIFF};
 
+    pub const EXTRA_DIFF: &'static [(&'static str,
+                                     &'static [&'static str],
+                                     &'static [&'static str])] = &[
+        // Official test suite doesn't include two Prepend chars between two other chars.
+        ("\u{20}\u{600}\u{600}\u{20}",
+         &["\u{20}", "\u{600}\u{600}\u{20}"],
+         &["\u{20}", "\u{600}", "\u{600}", "\u{20}"]),
+    ];
+
     for &(s, g) in TEST_SAME {
         // test forward iterator
         assert!(UnicodeSegmentation::graphemes(s, true)
@@ -34,11 +43,11 @@ fn test_graphemes() {
                 .all(|(a,b)| a == b));
     }
 
-    for &(s, gt, gf) in TEST_DIFF {
+    for &(s, gt, gf) in TEST_DIFF.iter().chain(EXTRA_DIFF) {
         // test forward iterator
         assert!(UnicodeSegmentation::graphemes(s, true)
                 .zip(gt.iter().cloned())
-                .all(|(a,b)| a == b));
+                .all(|(a,b)| a == b), "{:?}", s);
         assert!(UnicodeSegmentation::graphemes(s, false)
                 .zip(gf.iter().cloned())
                 .all(|(a,b)| a == b));