Merge pull request #5 from vndangkhoa/devin/1782475848-telex-spacing-clipboard

Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
Co-authored-by: vndangkhoa <vonguyendangkhoa@gmail.com>

daemon/src/main.rs

@@ -921,6 +921,7 @@ fn run_with_evdev(
                             if ch.is_ascii_alphabetic() && (shift ^ caps) {
                                 ch = ch.to_ascii_uppercase();
                             }
+                            let buf_before = daemon.engine.buffer().chars().count();
                             let commands = daemon.process_key(ch);
                             if !commands.is_empty() {
                                 consumed_keys.insert(keycode);
@@ -933,8 +934,15 @@ fn run_with_evdev(
                                 }
                                 // Skip upcoming auto-repeat pile-up from injection delay
                                 skip_count = 3;
-                            } else if is_vn_control_key(&daemon.config.input_method, ch) {
+                            } else if is_vn_control_key(&daemon.config.input_method, ch)
-                                // Tone/mark key with no effect — consume silently
+                                && daemon.engine.buffer().chars().count() <= buf_before
+                            {
+                                // Tone/mark key truly absorbed with no effect (no
+                                // literal character appended) — consume silently.
+                                // When the key is instead kept as a literal base
+                                // letter (e.g. leading "x", the "r" in "tr"), the
+                                // buffer grows and we must forward it like any
+                                // other character so it reaches the screen.
                                 consumed_keys.insert(keycode);
                             } else {
                                 injector.send_key_event(keycode, 1);
@@ -1250,3 +1258,99 @@ fn key_to_char(key: evdev::Key) -> Option<char> {
         _ => None,
     }
 }
+
+#[cfg(test)]
+mod grab_render_tests {
+    //! Models the grab-mode keystroke loop (the `value == 1` branch of
+    //! `run_with_evdev`) against a real engine, rendering the resulting
+    //! on-screen text. This exercises both the engine composition and the
+    //! daemon's decision of when to forward a raw key vs. consume it.
+    use super::*;
+
+    fn event_to_commands(event: Option<EngineEvent>) -> Vec<OutputCommand> {
+        let mut commands = Vec::new();
+        if let Some(event) = event {
+            match event {
+                EngineEvent::Flush(text) | EngineEvent::Insert(text) | EngineEvent::Paste(text) => {
+                    commands.push(OutputCommand::Type(text));
+                }
+                EngineEvent::AutoRestore(word) => {
+                    commands.push(OutputCommand::Backspace(word.chars().count()));
+                    commands.push(OutputCommand::Type(word));
+                }
+                EngineEvent::Replace { backspaces, insert } => {
+                    commands.push(OutputCommand::Backspace(backspaces));
+                    commands.push(OutputCommand::Type(insert));
+                }
+                EngineEvent::UndoTones { backspaces, restored } => {
+                    commands.push(OutputCommand::Backspace(backspaces));
+                    commands.push(OutputCommand::Type(restored));
+                }
+            }
+        }
+        commands
+    }
+
+    /// Render keystrokes exactly as the grab-mode loop would put them on screen.
+    fn render(method_str: &str, keys: &str) -> String {
+        let method = match method_str {
+            "vni" => InputMethod::Vni,
+            _ => InputMethod::Telex,
+        };
+        let mut engine = Engine::new(method);
+        engine.set_enabled(true);
+        engine.set_auto_restore(true);
+
+        let mut screen: Vec<char> = Vec::new();
+        for ch in keys.chars() {
+            let buf_before = engine.buffer().chars().count();
+            let commands = event_to_commands(engine.process_key(ch));
+            if !commands.is_empty() {
+                for cmd in &commands {
+                    match cmd {
+                        OutputCommand::Backspace(n) => {
+                            for _ in 0..*n {
+                                screen.pop();
+                            }
+                        }
+                        OutputCommand::Type(text) => screen.extend(text.chars()),
+                    }
+                }
+                if is_flush_char(ch) {
+                    screen.push(ch);
+                }
+            } else if is_vn_control_key(method_str, ch)
+                && engine.buffer().chars().count() <= buf_before
+            {
+                // consumed silently
+            } else {
+                screen.push(ch);
+            }
+        }
+        screen.into_iter().collect()
+    }
+
+    #[test]
+    fn leading_control_letters_are_kept() {
+        // "x" tone key as a leading consonant must survive.
+        assert_eq!(render("telex", "xuaw"), "xưa");
+        // "r" inside the "tr" initial cluster must not be eaten as a tone.
+        assert_eq!(render("telex", "trong"), "trong");
+        // "r" as a real word-initial consonant.
+        assert_eq!(render("telex", "ruwngf"), "rừng");
+    }
+
+    #[test]
+    fn spaces_between_words_are_preserved() {
+        assert_eq!(render("telex", "Ngayf xuaw"), "Ngày xưa");
+        assert_eq!(render("telex", "khu ruwngf raamj"), "khu rừng rậm");
+        assert_eq!(render("telex", "con Voi raats"), "con Voi rất");
+    }
+
+    #[test]
+    fn full_sentence_renders_correctly() {
+        let keys = "Ngayf xuaw, trong mootj khu ruwngf raamj cos mootj con Voi raats hung duwx.";
+        let expected = "Ngày xưa, trong một khu rừng rậm có một con Voi rất hung dữ.";
+        assert_eq!(render("telex", keys), expected);
+    }
+}

engine/src/bamboo.rs

@@ -127,6 +127,30 @@ impl BambooEngine {
                     }
                 }
             }
+
+            // Smart "ua" → "ưa": the horn goes on the u (xưa, chưa, mưa, lửa),
+            // not the breve on the a ("xuă" is not a valid syllable). Skip the
+            // "qu" glide case, where the u belongs to the initial consonant and
+            // the a takes the breve instead (quă → quăng).
+            if self.composition.len() >= 2 {
+                let a_idx = self.composition.len() - 1;
+                let u_idx = a_idx - 1;
+                let a_ch = self.composition[a_idx].base_char.to_ascii_lowercase();
+                let u_ch = self.composition[u_idx].base_char.to_ascii_lowercase();
+                let preceded_by_q = u_idx > 0
+                    && self.composition[u_idx - 1]
+                        .base_char
+                        .eq_ignore_ascii_case(&'q');
+                if a_ch == 'a'
+                    && u_ch == 'u'
+                    && self.composition[u_idx].mark_applied.is_none()
+                    && !preceded_by_q
+                {
+                    self.composition[u_idx].base_char = 'ư';
+                    self.composition[u_idx].mark_applied = Some('ư');
+                    return Some(self.flatten());
+                }
+            }
         }
 
         // Try mark rules with flexible backtrack" (scan up to 3 chars backward)
@@ -580,6 +604,23 @@ fn test_telex_gios() {
 
 
 
+    #[test]
+    fn test_telex_ua_horn() {
+        // "w" after a "ua" cluster puts the horn on the u (ưa), it must not
+        // put the breve on the a ("xuă" is not a valid Vietnamese syllable).
+        assert_eq!(process(InputMethod::Telex, "xuaw"), "xưa");
+        assert_eq!(process(InputMethod::Telex, "chuaw"), "chưa");
+        assert_eq!(process(InputMethod::Telex, "muaw"), "mưa");
+        assert_eq!(process(InputMethod::Telex, "Xuaw"), "Xưa");
+        // With a following tone the horn target still carries the tone.
+        assert_eq!(process(InputMethod::Telex, "luawr"), "lửa");
+        // "qu" glide exception: the u belongs to the initial, a takes the breve.
+        assert_eq!(process(InputMethod::Telex, "quawng"), "quăng");
+        // VNI parity.
+        assert_eq!(process(InputMethod::Vni, "xua7"), "xưa");
+        assert_eq!(process(InputMethod::Vni, "qua8ng"), "quăng");
+    }
+
     #[test]
     fn test_telex_r_as_normal_char() {
         let mut e = BambooEngine::new(InputMethod::Telex);

protocol/src/uinput_monitor.rs

@@ -18,6 +18,12 @@ const KEY_MAX: u32 = 0x1ff;
 
 pub struct UinputInjector {
     file: File,
+    /// The user's real clipboard contents, saved before we overwrite the
+    /// clipboard to inject Unicode text, so we can restore it afterwards.
+    saved_clipboard: std::sync::Mutex<Option<String>>,
+    /// The last text we injected via the clipboard. Used to tell our own
+    /// injected text apart from text the user copied with Ctrl+C.
+    last_injected: std::sync::Mutex<Option<String>>,
 }
 
 unsafe impl Send for UinputInjector {}
@@ -72,7 +78,11 @@ impl UinputInjector {
         // Small delay for device to be ready
         std::thread::sleep(std::time::Duration::from_millis(10));
 
-        Ok(Self { file })
+        Ok(Self {
+            file,
+            saved_clipboard: std::sync::Mutex::new(None),
+            last_injected: std::sync::Mutex::new(None),
+        })
     }
 
     fn send_uinput_event(&self, type_: u16, code: u16, value: i32) {
@@ -180,10 +190,8 @@ impl KeyInjector for UinputInjector {
             "[vietc] send_string: Unicode '{}' - using clipboard",
             s.escape_default()
         );
-        let copied = self.copy_to_clipboard(s);
+        let copied = self.paste_via_clipboard(s, false);
         if copied {
-            eprintln!("[vietc] send_string: clipboard OK, sending Ctrl+V");
-            self.send_ctrl_v();
             eprintln!("[vietc] send_string complete (clipboard)");
             return InjectResult::Success;
         } else {
@@ -365,13 +373,67 @@ impl UinputInjector {
         if backspaces > 0 {
             for _ in 0..backspaces { let _ = self.send_backspace(); }
         }
-        if self.copy_to_clipboard(text) {
+        self.paste_via_clipboard(text, true);
-            self.send_ctrl_v_x11();
-        }
 
         InjectResult::Success
     }
 
+    /// Read the user's current clipboard contents (wl-paste on Wayland, xclip
+    /// on X11). Returns None if no clipboard tool is available or it is empty.
+    fn read_clipboard(&self) -> Option<String> {
+        let is_wayland = std::env::var("WAYLAND_DISPLAY").is_ok();
+        let (prog, args): (&str, &[&str]) = if is_wayland {
+            ("wl-paste", &["-n"])
+        } else {
+            ("xclip", &["-selection", "clipboard", "-o"])
+        };
+        let mut cmd = Self::user_cmd(prog);
+        cmd.args(args);
+        let output = cmd.output().ok()?;
+        if !output.status.success() {
+            return None;
+        }
+        Some(String::from_utf8_lossy(&output.stdout).into_owned())
+    }
+
+    /// Inject Unicode `text` by placing it on the clipboard and sending Ctrl+V,
+    /// while preserving the user's own clipboard contents. Without this, every
+    /// Vietnamese word the user types would overwrite whatever they had copied
+    /// with Ctrl+C, so a subsequent Ctrl+V would paste the wrong thing.
+    ///
+    /// Returns whether the text was successfully copied to the clipboard.
+    fn paste_via_clipboard(&self, text: &str, use_x11_paste: bool) -> bool {
+        // Snapshot the clipboard. If it differs from what we last injected, the
+        // user changed it themselves (a real Ctrl+C), so remember it to restore.
+        let current = self.read_clipboard();
+        {
+            let last = self.last_injected.lock().unwrap();
+            let is_our_injection = matches!((&current, &*last), (Some(c), Some(l)) if c == l);
+            if !is_our_injection {
+                *self.saved_clipboard.lock().unwrap() = current;
+            }
+        }
+
+        if !self.copy_to_clipboard(text) {
+            return false;
+        }
+        if use_x11_paste {
+            self.send_ctrl_v_x11();
+        } else {
+            self.send_ctrl_v();
+        }
+
+        // Restore the user's clipboard once the paste has been consumed. The
+        // extra delay gives the target application time to read our text from
+        // the clipboard before we overwrite it again.
+        std::thread::sleep(std::time::Duration::from_millis(40));
+        let saved = self.saved_clipboard.lock().unwrap().clone();
+        let restored = saved.unwrap_or_default();
+        let _ = self.copy_to_clipboard(&restored);
+        *self.last_injected.lock().unwrap() = Some(restored);
+        true
+    }
+
     /// Copy text to clipboard and paste via Ctrl+V through our uinput device.
     /// Only used as a last resort if Wayland/X11 direct typing tools are unavailable.
     /// Tries xdotool first (X11/XWayland), then clipboard fallback.

uinputd/src/main.rs

@@ -47,6 +47,12 @@ struct input_id {
 
 struct UinputDevice {
     fd: i32,
+    /// The user's real clipboard contents, saved before we overwrite the
+    /// clipboard to paste Unicode text, so we can restore it afterwards.
+    saved_clipboard: std::sync::Mutex<Option<String>>,
+    /// The last text we injected via the clipboard, used to distinguish our
+    /// own paste content from text the user copied with Ctrl+C.
+    last_injected: std::sync::Mutex<Option<String>>,
 }
 
 impl UinputDevice {
@@ -83,7 +89,11 @@ impl UinputDevice {
         std::thread::sleep(std::time::Duration::from_millis(10));
 
         eprintln!("[vietc-uinputd] Device '{}' created", name);
-        Ok(Self { fd })
+        Ok(Self {
+            fd,
+            saved_clipboard: std::sync::Mutex::new(None),
+            last_injected: std::sync::Mutex::new(None),
+        })
     }
 
     fn send_event(&self, type_: u16, code: u16, value: i32) {
@@ -153,6 +163,19 @@ impl UinputDevice {
     }
 
     fn paste_unicode(&self, text: &str) {
+        // Save the user's clipboard before we clobber it, unless what is on the
+        // clipboard is our own previously-injected text. This keeps Ctrl+C /
+        // Ctrl+V working: every Vietnamese word is pasted via the clipboard, so
+        // without restoring it the user's copied content would be lost.
+        let current = read_clipboard();
+        {
+            let last = self.last_injected.lock().unwrap();
+            let is_our_injection = matches!((&current, &*last), (Some(c), Some(l)) if c == l);
+            if !is_our_injection {
+                *self.saved_clipboard.lock().unwrap() = current;
+            }
+        }
+
         copy_to_clipboard(text);
         self.send_key(29, 1);
         std::thread::sleep(std::time::Duration::from_millis(2));
@@ -160,6 +183,13 @@ impl UinputDevice {
         self.send_key(47, 0);
         self.send_key(29, 0);
         std::thread::sleep(std::time::Duration::from_millis(10));
+
+        // Restore the user's clipboard after the paste has been consumed.
+        std::thread::sleep(std::time::Duration::from_millis(30));
+        let saved = self.saved_clipboard.lock().unwrap().clone();
+        let restored = saved.unwrap_or_default();
+        copy_to_clipboard(&restored);
+        *self.last_injected.lock().unwrap() = Some(restored);
     }
 }
 
@@ -171,6 +201,22 @@ impl Drop for UinputDevice {
     }
 }
 
+fn read_clipboard() -> Option<String> {
+    let is_wayland = std::env::var("WAYLAND_DISPLAY").is_ok();
+    let output = if is_wayland {
+        Command::new("wl-paste").arg("-n").output()
+    } else {
+        Command::new("xclip")
+            .args(["-selection", "clipboard", "-o"])
+            .output()
+    };
+    let output = output.ok()?;
+    if !output.status.success() {
+        return None;
+    }
+    Some(String::from_utf8_lossy(&output.stdout).into_owned())
+}
+
 fn copy_to_clipboard(text: &str) {
     let is_wayland = std::env::var("WAYLAND_DISPLAY").is_ok();
     if is_wayland {