Preserve grapheme clusters when shortening desktop labels

crates/oa-desktop/src/views.rs

@@ -12,6 +12,7 @@ use iced::{
     },
 };
 use oa_core::*;
+use unicode_segmentation::UnicodeSegmentation;
 
 pub fn label<'a>(s: impl Into<String>) -> widget::Text<'a> {
     text(s.into()).size(11).color(style::MUTED)
@@ -24,10 +25,12 @@ pub fn action<'a>(s: &str, m: Message) -> widget::Button<'a, Message> {
 }
 pub fn short(s: &str, limit: usize) -> String {
     let s = s.split_whitespace().collect::<Vec<_>>().join(" ");
-    if s.chars().count() > limit {
+    if s.graphemes(true).count() > limit {
         format!(
             "{}…",
-            s.chars().take(limit.saturating_sub(1)).collect::<String>()
+            s.graphemes(true)
+                .take(limit.saturating_sub(1))
+                .collect::<String>()
         )
     } else {
         s
@@ -1290,3 +1293,57 @@ fn checks(app: &App) -> View<'_, Message> {
     }
     scrollable(out.padding(24)).into()
 }
+
+#[cfg(test)]
+mod tests {
+    use super::short;
+
+    #[test]
+    fn short_preserves_native_lifeline_clusters_at_cutoff() {
+        let oa = "abcdefghijklmnope\u{0301}尾";
+        assert_eq!(short(oa, 17), "abcdefghijklmnop…");
+        assert_eq!(short(oa, 18), oa);
+        assert_eq!(short(oa, 19), oa);
+
+        let sa = "abcdefghijklmnop👩\u{200d}🔬 tail";
+        assert_eq!(short(sa, 17), "abcdefghijklmnop…");
+        assert_eq!(short(sa, 18), "abcdefghijklmnop👩\u{200d}🔬…");
+        assert_eq!(short(sa, 19), "abcdefghijklmnop👩\u{200d}🔬 …");
+        assert_eq!(short(sa, 22), sa);
+    }
+
+    #[test]
+    fn short_never_splits_retained_extended_graphemes() {
+        for cluster in ["e\u{0301}", "👩\u{200d}🔬", "👍🏽", "🇫🇷"] {
+            let value = format!("A{cluster}BC");
+            assert_eq!(short(&value, 2), "A…");
+            assert_eq!(short(&value, 3), format!("A{cluster}…"));
+            assert_eq!(short(&value, 4), value);
+        }
+    }
+
+    #[test]
+    fn short_preserves_ascii_cjk_and_whitespace_policy() {
+        assert_eq!(short("abcd", 3), "ab…");
+        assert_eq!(short("abcd", 4), "abcd");
+        assert_eq!(short("abcd", 5), "abcd");
+        assert_eq!(short("甲乙丙丁", 3), "甲乙…");
+        assert_eq!(short("甲乙丙丁", 4), "甲乙丙丁");
+        assert_eq!(short(" \tA\r\n e\u{0301}\u{2003}B  ", 5), "A e\u{0301} B");
+        assert_eq!(short(" \tA\r\n e\u{0301}\u{2003}B  ", 4), "A e\u{0301}…");
+    }
+
+    #[test]
+    fn short_preserves_empty_and_small_limit_policy() {
+        for limit in [0, 1, 2] {
+            assert_eq!(short("", limit), "");
+            assert_eq!(short(" \t\r\n", limit), "");
+        }
+        assert_eq!(short("A", 0), "…");
+        assert_eq!(short("A", 1), "A");
+        assert_eq!(short("AB", 1), "…");
+        assert_eq!(short("e\u{0301}", 0), "…");
+        assert_eq!(short("e\u{0301}", 1), "e\u{0301}");
+        assert_eq!(short("e\u{0301}B", 1), "…");
+    }
+}