voice transcription (#3381)

Adds voice transcription on press-and-hold of spacebar.


https://github.com/user-attachments/assets/85039314-26f3-46d1-a83b-8c4a4a1ecc21

---------

Co-authored-by: Codex <199175422+chatgpt-codex-connector[bot]@users.noreply.github.com>
Co-authored-by: David Zbarsky <zbarsky@openai.com>

codex-rs/tui/src/bottom_pane/mod.rs

@@ -33,6 +33,7 @@ use codex_protocol::request_user_input::RequestUserInputEvent;
 use codex_protocol::user_input::TextElement;
 use crossterm::event::KeyCode;
 use crossterm::event::KeyEvent;
+use crossterm::event::KeyEventKind;
 use ratatui::buffer::Buffer;
 use ratatui::layout::Rect;
 use ratatui::text::Line;
@@ -204,8 +205,8 @@ impl BottomPane {
             placeholder_text,
             disable_paste_burst,
         );
+        composer.set_frame_requester(frame_requester.clone());
         composer.set_skill_mentions(skills);
-
         Self {
             composer,
             view_stack: Vec::new(),
@@ -291,6 +292,11 @@ impl BottomPane {
         self.request_redraw();
     }
 
+    pub fn set_voice_transcription_enabled(&mut self, enabled: bool) {
+        self.composer.set_voice_transcription_enabled(enabled);
+        self.request_redraw();
+    }
+
     /// Update the key hint shown next to queued messages so it matches the
     /// binding that `ChatWidget` actually listens for.
     pub(crate) fn set_queued_message_edit_binding(&mut self, binding: KeyBinding) {
@@ -327,8 +333,23 @@ impl BottomPane {
 
     /// Forward a key event to the active view or the composer.
     pub fn handle_key_event(&mut self, key_event: KeyEvent) -> InputResult {
+        // Do not globally intercept space; only composer handles hold-to-talk.
+        // While recording, route all keys to the composer so it can stop on release or next key.
+        #[cfg(not(target_os = "linux"))]
+        if self.composer.is_recording() {
+            let (_ir, needs_redraw) = self.composer.handle_key_event(key_event);
+            if needs_redraw {
+                self.request_redraw();
+            }
+            return InputResult::None;
+        }
+
         // If a modal/view is active, handle it here; otherwise forward to composer.
         if !self.view_stack.is_empty() {
+            if key_event.kind == KeyEventKind::Release {
+                return InputResult::None;
+            }
+
             // We need three pieces of information after routing the key:
             // whether Esc completed the view, whether the view finished for any
             // reason, and whether a paste-burst timer should be scheduled.
@@ -432,6 +453,7 @@ impl BottomPane {
             }
         } else {
             let needs_redraw = self.composer.handle_paste(pasted);
+            self.composer.sync_popups();
             if needs_redraw {
                 self.request_redraw();
             }
@@ -440,9 +462,18 @@ impl BottomPane {
 
     pub(crate) fn insert_str(&mut self, text: &str) {
         self.composer.insert_str(text);
+        self.composer.sync_popups();
         self.request_redraw();
     }
 
+    // Space hold timeout is handled inside ChatComposer via an internal timer.
+    pub(crate) fn pre_draw_tick(&mut self) {
+        // Allow composer to process any time-based transitions before drawing
+        #[cfg(not(target_os = "linux"))]
+        self.composer.process_space_hold_trigger();
+        self.composer.sync_popups();
+    }
+
     /// Replace the composer text with `text`.
     ///
     /// This is intended for fresh input where mention linkage does not need to
@@ -895,6 +926,7 @@ impl BottomPane {
             .on_history_entry_response(log_id, offset, entry);
 
         if updated {
+            self.composer.sync_popups();
             self.request_redraw();
         }
     }
@@ -973,6 +1005,30 @@ impl BottomPane {
     }
 }
 
+#[cfg(not(target_os = "linux"))]
+impl BottomPane {
+    pub(crate) fn replace_transcription(&mut self, id: &str, text: &str) {
+        self.composer.replace_transcription(id, text);
+        self.composer.sync_popups();
+        self.request_redraw();
+    }
+
+    pub(crate) fn update_transcription_in_place(&mut self, id: &str, text: &str) -> bool {
+        let updated = self.composer.update_transcription_in_place(id, text);
+        if updated {
+            self.composer.sync_popups();
+            self.request_redraw();
+        }
+        updated
+    }
+
+    pub(crate) fn remove_transcription_placeholder(&mut self, id: &str) {
+        self.composer.remove_transcription_placeholder(id);
+        self.composer.sync_popups();
+        self.request_redraw();
+    }
+}
+
 impl Renderable for BottomPane {
     fn render(&self, area: Rect, buf: &mut Buffer) {
         self.as_renderable().render(area, buf);
@@ -993,6 +1049,7 @@ mod tests {
     use crate::status_indicator_widget::StatusDetailsCapitalization;
     use codex_protocol::protocol::Op;
     use codex_protocol::protocol::SkillScope;
+    use crossterm::event::KeyEventKind;
     use crossterm::event::KeyModifiers;
     use insta::assert_snapshot;
     use ratatui::buffer::Buffer;
@@ -1571,4 +1628,58 @@ mod tests {
         assert_eq!(on_ctrl_c_calls.get(), 0);
         assert_eq!(handle_calls.get(), 1);
     }
+
+    #[test]
+    fn release_events_are_ignored_for_active_view() {
+        #[derive(Default)]
+        struct CountingView {
+            handle_calls: Rc<Cell<usize>>,
+        }
+
+        impl Renderable for CountingView {
+            fn render(&self, _area: Rect, _buf: &mut Buffer) {}
+
+            fn desired_height(&self, _width: u16) -> u16 {
+                0
+            }
+        }
+
+        impl BottomPaneView for CountingView {
+            fn handle_key_event(&mut self, _key_event: KeyEvent) {
+                self.handle_calls
+                    .set(self.handle_calls.get().saturating_add(1));
+            }
+        }
+
+        let (tx_raw, _rx) = unbounded_channel::<AppEvent>();
+        let tx = AppEventSender::new(tx_raw);
+        let mut pane = BottomPane::new(BottomPaneParams {
+            app_event_tx: tx,
+            frame_requester: FrameRequester::test_dummy(),
+            has_input_focus: true,
+            enhanced_keys_supported: false,
+            placeholder_text: "Ask Codex to do anything".to_string(),
+            disable_paste_burst: false,
+            animations_enabled: true,
+            skills: Some(Vec::new()),
+        });
+
+        let handle_calls = Rc::new(Cell::new(0));
+        pane.push_view(Box::new(CountingView {
+            handle_calls: Rc::clone(&handle_calls),
+        }));
+
+        pane.handle_key_event(KeyEvent::new_with_kind(
+            KeyCode::Down,
+            KeyModifiers::NONE,
+            KeyEventKind::Press,
+        ));
+        pane.handle_key_event(KeyEvent::new_with_kind(
+            KeyCode::Down,
+            KeyModifiers::NONE,
+            KeyEventKind::Release,
+        ));
+
+        assert_eq!(handle_calls.get(), 1);
+    }
 }