Reveal streaming text by grapheme

Author: DinahK-2SO

tools/wta/Cargo.lock

@@ -22,6 +22,7 @@ crossterm = { version = "0.29", features = ["event-stream"] }
 notify = "6"
 futures = "0.3"
 unicode-width = "0.2"
+unicode-segmentation = "1"
 textwrap = "0.16"
 tui-markdown = { version = "0.3.9", default-features = false }
 serde = { version = "1", features = ["derive"] }

tools/wta/Cargo.toml

@@ -8,6 +8,7 @@ use std::io;
 use std::sync::atomic::{AtomicBool, Ordering};
 use std::sync::Arc;
 use tokio::sync::mpsc;
+use unicode_segmentation::UnicodeSegmentation;
 
 struct DeferredAcpParams {
     agent_cmd: String,
@@ -4386,10 +4387,10 @@ impl App {
         }
     }
 
-    /// Number of user-visible characters in the active assistant segment.
+    /// Number of user-visible grapheme clusters in the active assistant segment.
     fn tab_visible_stream_len(tab: &TabSession) -> Option<usize> {
         crate::ui::chat::user_visible_stream_text(tab.streaming_agent_text()?)
-            .map(|text| text.chars().count())
+            .map(|text| text.graphemes(true).count())
     }
 
     /// True iff the current (visible) tab has streaming text that the reveal

tools/wta/src/app.rs

@@ -451,8 +451,8 @@ pub struct TabSession {
     // (see `doc/specs/turn-state-refactor.md`).
     pub turn: TurnState,
     pub activity_frame: usize,
-    /// Typewriter reveal cursor for the final assistant-text item in the
-    /// active transcript. Advanced toward its full length by `RevealTick`
+    /// Typewriter reveal cursor, in extended grapheme clusters, for the final
+    /// assistant-text item in the active transcript. Advanced toward its full length by `RevealTick`
     /// (`advance_reveal`), reset to 0 when a new turn starts streaming, and
     /// made irrelevant on finalize (the committed message renders in full).
     pub reveal_chars: usize,

tools/wta/src/app/tab_state.rs

@@ -8654,6 +8654,21 @@ fn first_message_chunk_transitions_to_streaming_with_transcript_text() {
     );
 }
 
+#[test]
+fn reveal_backlog_and_cursor_count_extended_grapheme_clusters() {
+    let mut app = test_app();
+    submit_test_prompt(&mut app, "hi");
+    app.turn_observe_chunk(DEFAULT_TAB_ID, ChunkKind::Message, "👩‍💻x");
+
+    assert_eq!(App::tab_visible_stream_len(app.current_tab()), Some(2));
+    app.current_tab_mut().reveal_chars = 1;
+    assert!(app.has_reveal_backlog());
+
+    app.advance_reveal();
+    assert_eq!(app.current_tab().reveal_chars, 2);
+    assert!(!app.has_reveal_backlog());
+}
+
 #[test]
 fn thought_chunk_first_transitions_without_visible_text() {
     let mut app = test_app();

tools/wta/src/app_tests.rs

@@ -8,6 +8,7 @@ use std::rc::Rc;
 use ratatui::prelude::*;
 use ratatui::widgets::{Block, Borders, Paragraph, Wrap};
 use tui_markdown::{Options as MarkdownOptions, StyleSheet};
+use unicode_segmentation::UnicodeSegmentation;
 use unicode_width::UnicodeWidthChar;
 use unicode_width::UnicodeWidthStr;
 
@@ -1357,19 +1358,19 @@ fn build_pending_stream_lines_for_tab_with_mode<'a>(
     let Some(text) = pending_render_text(tab) else {
         return Vec::new();
     };
-    // Typewriter smoothing: only reveal the first `reveal_chars` characters of
-    // the streaming text. The reveal cursor is advanced toward the full length
+    // Typewriter smoothing: only reveal the first `reveal_chars` grapheme
+    // clusters of the streaming text. The reveal cursor advances toward the full length
     // by the `RevealTick` animation (`App::advance_reveal`), turning the
     // upstream ~90-char-every-~100ms bursts into a smooth character flow. The
     // full text is always in the ordered transcript, and finalize moves that
     // transcript to history unchanged.
     let revealed: Cow<'_, str> = {
-        let total = text.chars().count();
+        let total = text.graphemes(true).count();
         let shown = tab.reveal_chars.max(1).min(total);
         if shown >= total {
             text
         } else {
-            Cow::Owned(text.chars().take(shown).collect())
+            Cow::Owned(text.graphemes(true).take(shown).collect())
         }
     };
     agent_response_lines(
@@ -2467,6 +2468,20 @@ mod tests {
         assert!(partial_text.contains("bo"));
     }
 
+    #[test]
+    fn pending_stream_reveal_does_not_split_grapheme_clusters() {
+        let text = "👩‍💻 ready";
+        let tab = streaming_tab(text, 1);
+
+        let rendered = build_pending_stream_lines_for_tab(&tab, 80)
+            .iter()
+            .map(line_text)
+            .collect::<Vec<_>>()
+            .join("\n");
+
+        assert_eq!(rendered, "● 👩‍💻");
+    }
+
     #[test]
     fn thinking_activity_follows_turn_lifecycle() {
         let mut tab = streaming_tab("", 0);