test: add ime input regressions

8 files changed, 201 insertions, 17 deletions

diff --git a/src/app/mod.rs b/src/app/mod.rs
index 43d15bec..7534559f 100644
--- a/src/app/mod.rs
+++ b/src/app/mod.rs
@@ -1979,6 +1979,56 @@ mod tests {
         assert!(rx.try_recv().is_err());
     }
 
+    #[tokio::test]
+    async fn route_client_input_forwards_multilingual_ime_text_to_focused_pane() {
+        let mut app = test_app();
+        let mut workspace = Workspace::test_new("test");
+        let focused = workspace.focused_pane_id().unwrap();
+        let text = "中日한🙂";
+        let (runtime, mut rx) =
+            PaneRuntime::test_with_channel_capacity(80, 24, text.chars().count());
+        workspace.tabs[0].runtimes.insert(focused, runtime);
+        app.state.workspaces = vec![workspace];
+        app.state.active = Some(0);
+        app.state.selected = 0;
+        app.state.mode = Mode::Terminal;
+
+        app.route_client_input(text.as_bytes().to_vec());
+
+        let mut forwarded = Vec::new();
+        for _ in text.chars() {
+            let chunk = rx.recv().await.unwrap();
+            forwarded.extend_from_slice(&chunk);
+        }
+        assert_eq!(forwarded, text.as_bytes());
+        assert!(rx.try_recv().is_err());
+    }
+
+    #[tokio::test]
+    async fn route_client_input_forwards_long_voice_like_cjk_text_without_truncation() {
+        let mut app = test_app();
+        let mut workspace = Workspace::test_new("test");
+        let focused = workspace.focused_pane_id().unwrap();
+        let text = "你好，今天我们测试一段比较长的语音输入。こんにちは。안녕하세요.🙂".repeat(64);
+        let char_count = text.chars().count();
+        let (runtime, mut rx) = PaneRuntime::test_with_channel_capacity(80, 24, char_count);
+        workspace.tabs[0].runtimes.insert(focused, runtime);
+        app.state.workspaces = vec![workspace];
+        app.state.active = Some(0);
+        app.state.selected = 0;
+        app.state.mode = Mode::Terminal;
+
+        app.route_client_input(text.as_bytes().to_vec());
+
+        let mut forwarded = Vec::new();
+        for _ in 0..char_count {
+            let chunk = rx.recv().await.unwrap();
+            forwarded.extend_from_slice(&chunk);
+        }
+        assert_eq!(forwarded, text.as_bytes());
+        assert!(rx.try_recv().is_err());
+    }
+
     #[test]
     fn route_client_input_handles_mouse_events() {
         let mut app = test_app();
diff --git a/src/client/mod.rs b/src/client/mod.rs
index 475a079f..e378427f 100644
--- a/src/client/mod.rs
+++ b/src/client/mod.rs
@@ -23,7 +23,7 @@ use std::time::Duration;
 
 use crossterm::event::{
     DisableBracketedPaste, DisableFocusChange, DisableMouseCapture, EnableBracketedPaste,
-    EnableFocusChange, EnableMouseCapture, KeyboardEnhancementFlags, PopKeyboardEnhancementFlags,
+    EnableFocusChange, EnableMouseCapture, PopKeyboardEnhancementFlags,
     PushKeyboardEnhancementFlags,
 };
 use crossterm::execute;
@@ -146,11 +146,7 @@ fn setup_terminal() -> io::Result<TerminalGuard> {
         EnableMouseCapture,
         EnableBracketedPaste,
         EnableFocusChange,
-        PushKeyboardEnhancementFlags(
-            KeyboardEnhancementFlags::DISAMBIGUATE_ESCAPE_CODES
-                | KeyboardEnhancementFlags::REPORT_EVENT_TYPES
-                | KeyboardEnhancementFlags::REPORT_ALTERNATE_KEYS
-        )
+        PushKeyboardEnhancementFlags(crate::input::ime_compatible_keyboard_enhancement_flags())
     )?;
 
     // tmux doesn't understand kitty keyboard protocol push.
diff --git a/src/input/mod.rs b/src/input/mod.rs
index a2775e40..b4886256 100644
--- a/src/input/mod.rs
+++ b/src/input/mod.rs
@@ -6,5 +6,8 @@ mod parse;
 pub use encode::{
     encode_cursor_key, encode_mouse_button, encode_mouse_scroll, encode_terminal_key,
 };
-pub use model::{KeyboardProtocol, MouseProtocolEncoding, MouseProtocolMode, TerminalKey};
+pub use model::{
+    ime_compatible_keyboard_enhancement_flags, KeyboardProtocol, MouseProtocolEncoding,
+    MouseProtocolMode, TerminalKey,
+};
 pub use parse::parse_terminal_key_sequence;
diff --git a/src/input/model.rs b/src/input/model.rs
index 181ba7b3..95732bbc 100644
--- a/src/input/model.rs
+++ b/src/input/model.rs
@@ -1,4 +1,4 @@
-use crossterm::event::{KeyCode, KeyEvent, KeyModifiers};
+use crossterm::event::{KeyCode, KeyEvent, KeyModifiers, KeyboardEnhancementFlags};
 
 #[derive(Debug, Clone, Copy, PartialEq, Eq)]
 pub struct TerminalKey {
@@ -40,6 +40,12 @@ impl From<KeyEvent> for TerminalKey {
     }
 }
 
+pub fn ime_compatible_keyboard_enhancement_flags() -> KeyboardEnhancementFlags {
+    KeyboardEnhancementFlags::DISAMBIGUATE_ESCAPE_CODES
+        | KeyboardEnhancementFlags::REPORT_EVENT_TYPES
+        | KeyboardEnhancementFlags::REPORT_ALTERNATE_KEYS
+}
+
 #[derive(Debug, Clone, Copy, PartialEq, Eq)]
 pub enum KeyboardProtocol {
     Legacy,
@@ -97,4 +103,14 @@ mod tests {
             KeyboardProtocol::Kitty { flags: 7 }
         );
     }
+
+    #[test]
+    fn keyboard_enhancement_flags_stay_ime_compatible() {
+        let flags = ime_compatible_keyboard_enhancement_flags();
+
+        assert!(flags.contains(KeyboardEnhancementFlags::DISAMBIGUATE_ESCAPE_CODES));
+        assert!(flags.contains(KeyboardEnhancementFlags::REPORT_EVENT_TYPES));
+        assert!(flags.contains(KeyboardEnhancementFlags::REPORT_ALTERNATE_KEYS));
+        assert!(!flags.contains(KeyboardEnhancementFlags::REPORT_ALL_KEYS_AS_ESCAPE_CODES));
+    }
 }
diff --git a/src/main.rs b/src/main.rs
index f4588bdc..dceb1230 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -2,7 +2,7 @@ use std::io;
 
 use crossterm::event::{
     DisableBracketedPaste, DisableFocusChange, DisableMouseCapture, EnableBracketedPaste,
-    EnableFocusChange, EnableMouseCapture, KeyboardEnhancementFlags, PopKeyboardEnhancementFlags,
+    EnableFocusChange, EnableMouseCapture, PopKeyboardEnhancementFlags,
     PushKeyboardEnhancementFlags,
 };
 use crossterm::execute;
@@ -454,11 +454,7 @@ fn main() -> io::Result<()> {
             EnableMouseCapture,
             EnableBracketedPaste,
             EnableFocusChange,
-            PushKeyboardEnhancementFlags(
-                KeyboardEnhancementFlags::DISAMBIGUATE_ESCAPE_CODES
-                    | KeyboardEnhancementFlags::REPORT_EVENT_TYPES
-                    | KeyboardEnhancementFlags::REPORT_ALTERNATE_KEYS
-            )
+            PushKeyboardEnhancementFlags(crate::input::ime_compatible_keyboard_enhancement_flags())
         )?;
 
         // tmux doesn't understand kitty keyboard protocol push (\e[>1u).
diff --git a/src/pane.rs b/src/pane.rs
index bf669ecf..98edf403 100644
--- a/src/pane.rs
+++ b/src/pane.rs
@@ -925,7 +925,15 @@ impl PaneRuntime {
 #[cfg(test)]
 impl PaneRuntime {
     pub(crate) fn test_with_channel(cols: u16, rows: u16) -> (Self, mpsc::Receiver<Bytes>) {
-        Self::test_with_channel_and_scrollback_bytes(cols, rows, 0, &[])
+        Self::test_with_channel_and_scrollback_bytes(cols, rows, 0, &[], 4)
+    }
+
+    pub(crate) fn test_with_channel_capacity(
+        cols: u16,
+        rows: u16,
+        capacity: usize,
+    ) -> (Self, mpsc::Receiver<Bytes>) {
+        Self::test_with_channel_and_scrollback_bytes(cols, rows, 0, &[], capacity)
     }
 
     pub(crate) fn test_with_screen_bytes(cols: u16, rows: u16, bytes: &[u8]) -> Self {
@@ -938,7 +946,7 @@ impl PaneRuntime {
         scrollback_limit_bytes: usize,
         bytes: &[u8],
     ) -> Self {
-        Self::test_with_channel_and_scrollback_bytes(cols, rows, scrollback_limit_bytes, bytes).0
+        Self::test_with_channel_and_scrollback_bytes(cols, rows, scrollback_limit_bytes, bytes, 4).0
     }
 
     fn test_with_channel_and_scrollback_bytes(
@@ -946,8 +954,9 @@ impl PaneRuntime {
         rows: u16,
         scrollback_limit_bytes: usize,
         bytes: &[u8],
+        channel_capacity: usize,
     ) -> (Self, mpsc::Receiver<Bytes>) {
-        let (tx, rx) = mpsc::channel(4);
+        let (tx, rx) = mpsc::channel(channel_capacity);
         let (resize_tx, _resize_rx) = watch::channel((rows, cols));
         let mut terminal =
             crate::ghostty::Terminal::new(cols, rows, scrollback_limit_bytes).unwrap();
diff --git a/src/raw_input.rs b/src/raw_input.rs
index 0cf9d84b..aa231595 100644
--- a/src/raw_input.rs
+++ b/src/raw_input.rs
@@ -1007,6 +1007,103 @@ mod tests {
     }
 
     #[test]
+    fn chunked_cjk_utf8_waits_for_all_continuation_bytes() {
+        let (tx, mut rx) = mpsc::channel(8);
+        let mut buffer = Vec::new();
+        let bytes = "好".as_bytes();
+
+        drain_chunk(&mut buffer, &tx, &bytes[..1]);
+        assert_eq!(buffer, bytes[..1]);
+        assert!(collect_events(&mut rx).is_empty());
+        flush_incomplete_buffer(&mut buffer, &tx);
+        assert_eq!(buffer, bytes[..1]);
+        assert!(collect_events(&mut rx).is_empty());
+
+        drain_chunk(&mut buffer, &tx, &bytes[1..2]);
+        assert_eq!(buffer, bytes[..2]);
+        assert!(collect_events(&mut rx).is_empty());
+        flush_incomplete_buffer(&mut buffer, &tx);
+        assert_eq!(buffer, bytes[..2]);
+        assert!(collect_events(&mut rx).is_empty());
+
+        drain_chunk(&mut buffer, &tx, &bytes[2..]);
+        assert!(buffer.is_empty());
+        let events = collect_events(&mut rx);
+        assert_eq!(events.len(), 1);
+        assert_raw_key(
+            events.into_iter().next().unwrap(),
+            KeyCode::Char('好'),
+            KeyModifiers::empty(),
+        );
+    }
+
+    #[test]
+    fn chunked_four_byte_utf8_waits_for_all_continuation_bytes() {
+        let (tx, mut rx) = mpsc::channel(8);
+        let mut buffer = Vec::new();
+        let bytes = "🙂".as_bytes();
+
+        for split in 1..bytes.len() {
+            drain_chunk(&mut buffer, &tx, &bytes[split - 1..split]);
+            assert_eq!(buffer, bytes[..split]);
+            assert!(collect_events(&mut rx).is_empty());
+            flush_incomplete_buffer(&mut buffer, &tx);
+            assert_eq!(buffer, bytes[..split]);
+            assert!(collect_events(&mut rx).is_empty());
+        }
+
+        drain_chunk(&mut buffer, &tx, &bytes[bytes.len() - 1..]);
+        assert!(buffer.is_empty());
+        let events = collect_events(&mut rx);
+        assert_eq!(events.len(), 1);
+        assert_raw_key(
+            events.into_iter().next().unwrap(),
+            KeyCode::Char('🙂'),
+            KeyModifiers::empty(),
+        );
+    }
+
+    #[test]
+    fn long_multilingual_voice_like_burst_drains_without_truncation() {
+        let text = "你好，今天我们测试一段比较长的语音输入。こんにちは。안녕하세요.🙂".repeat(128);
+        assert!(
+            text.len() > 4096,
+            "test input should exceed the client read buffer"
+        );
+        let mut buffer = text.as_bytes().to_vec();
+
+        let chunks = drain_complete_input_bytes(&mut buffer);
+        let rebuilt: Vec<u8> = chunks.into_iter().flatten().collect();
+
+        assert!(buffer.is_empty());
+        assert_eq!(rebuilt, text.as_bytes());
+    }
+
+    #[test]
+    fn long_multilingual_burst_survives_one_byte_chunks_and_timeouts() {
+        let text = "中文かなカナ한글🙂，。".repeat(64);
+        let mut buffer = Vec::new();
+        let mut rebuilt = Vec::new();
+
+        for byte in text.as_bytes() {
+            buffer.push(*byte);
+            for chunk in drain_complete_input_bytes(&mut buffer) {
+                rebuilt.extend(chunk);
+            }
+            if !buffer.is_empty() {
+                assert_eq!(flush_incomplete_input_bytes(&mut buffer), None);
+            }
+        }
+
+        for chunk in drain_complete_input_bytes(&mut buffer) {
+            rebuilt.extend(chunk);
+        }
+
+        assert!(buffer.is_empty());
+        assert_eq!(rebuilt, text.as_bytes());
+    }
+
+    #[test]
     fn parse_with_ranges_tracks_byte_offsets() {
         use super::parse_raw_input_bytes_with_ranges;
 
diff --git a/src/server/protocol.rs b/src/server/protocol.rs
index 15aa3784..5eda0231 100644
--- a/src/server/protocol.rs
+++ b/src/server/protocol.rs
@@ -566,6 +566,23 @@ mod tests {
     }
 
     #[test]
+    fn client_input_large_multilingual_payload_roundtrip() {
+        let text = "你好，今天我们测试一段比较长的语音输入。こんにちは。안녕하세요.🙂".repeat(1024);
+        assert!(text.len() > 64 * 1024);
+        assert!(text.len() < MAX_FRAME_SIZE);
+        let msg = ClientMessage::Input {
+            data: text.as_bytes().to_vec(),
+        };
+
+        let encoded = bincode::serde::encode_to_vec(&msg, bincode::config::standard()).unwrap();
+        let (decoded, consumed): (ClientMessage, _) =
+            bincode::serde::decode_from_slice(&encoded, bincode::config::standard()).unwrap();
+
+        assert_eq!(consumed, encoded.len());
+        assert_eq!(decoded, msg);
+    }
+
+    #[test]
     fn client_resize_roundtrip() {
         let msg = ClientMessage::Resize { cols: 80, rows: 24 };
         let encoded = bincode::serde::encode_to_vec(&msg, bincode::config::standard()).unwrap();