Merge pull request #1 from vndangkhoa/devin/1782468642-tone-placement

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

engine/src/bamboo.rs

@@ -288,6 +288,7 @@ impl BambooEngine {
     }
 
     fn find_tone_position(&self, range: std::ops::Range<usize>) -> usize {
+        let start = range.start;
         let mut vowels: Vec<usize> = Vec::new();
 
         for i in range {
@@ -301,6 +302,18 @@ impl BambooEngine {
             return self.composition.len().saturating_sub(1);
         }
 
+        // Exclude onset glides: in "qu…" the u and in "gi…" the i belong to the
+        // initial consonant, not the vowel nucleus — so they must never carry the
+        // tone (e.g. "quả" not "qủa", "giờ" not "gìơ"). Only strip the glide when
+        // another vowel follows it; bare "gì"/"qu" keep the letter as the nucleus.
+        if vowels.len() >= 2 && vowels[0] == start + 1 {
+            let onset = self.composition[start].base_char.to_ascii_lowercase();
+            let glide = self.composition[start + 1].base_char.to_ascii_lowercase();
+            if (onset == 'q' && glide == 'u') || (onset == 'g' && glide == 'i') {
+                vowels.remove(0);
+            }
+        }
+
         if vowels.len() == 1 {
             return vowels[0];
         }
@@ -322,7 +335,7 @@ impl BambooEngine {
         let tone_on_second = matches!((cv1, cv2),
             ('o', 'a') | ('o', 'e') | ('u', 'y') |
             ('i', 'ê') | ('y', 'ê') | ('u', 'ô') | ('ư', 'ơ') |
-            ('i', 'o') | ('u', 'â')
+            ('i', 'o') | ('u', 'â') | ('u', 'ê') | ('u', 'ơ')
         );
 
         if tone_on_second {

engine/tests/tone_placement.rs

@@ -0,0 +1,53 @@
+//! Regression tests for tone placement on syllables whose onset contains a
+//! glide letter ("qu", "gi") and on the "uê"/"uơ" vowel clusters.
+
+use std::collections::HashMap;
+use vietc_engine::{Engine, InputMethod};
+
+fn telex(keys: &str) -> String {
+    Engine::replay_keystrokes(InputMethod::Telex, &HashMap::new(), &keys.chars().collect::<Vec<_>>()).0
+}
+
+fn vni(keys: &str) -> String {
+    Engine::replay_keystrokes(InputMethod::Vni, &HashMap::new(), &keys.chars().collect::<Vec<_>>()).0
+}
+
+/// (telex keystrokes, vni keystrokes, expected word)
+const CASES: &[(&str, &str, &str)] = &[
+    // "qu" onset: the u is part of the consonant, tone stays on the nucleus.
+    ("quar", "qua3", "quả"),
+    ("quaf", "qua2", "quà"),
+    ("quas", "qua1", "quá"),
+    ("quaj", "qua5", "quạ"),
+    // "gi" onset: the i is part of the consonant, tone stays on the nucleus.
+    ("gias", "gia1", "giá"),
+    ("giof", "gio2", "giò"),
+    ("giowf", "gio72", "giờ"),
+    ("giups", "giu1p", "giúp"),
+    ("gieets", "gie61t", "giết"),
+    ("giuwowngf", "giuo7ng2", "giường"),
+    // "uê"/"uơ" clusters: tone belongs on the second vowel.
+    ("thuees", "thue61", "thuế"),
+    ("hueej", "hue65", "huệ"),
+    // Controls that must keep working: bare "gì", "uy", "uâ", "uô".
+    ("gif", "gi2", "gì"),
+    ("quys", "quy1", "quý"),
+    ("quaanf", "qua62n", "quần"),
+    ("muoons", "muo61n", "muốn"),
+];
+
+#[test]
+fn onset_glide_and_cluster_tone_placement() {
+    let mut fails = Vec::new();
+    for &(tk, vk, want) in CASES {
+        let gt = telex(tk);
+        if gt != want {
+            fails.push(format!("TELEX {tk:>10} -> {gt:>8}  want {want}"));
+        }
+        let gv = vni(vk);
+        if gv != want {
+            fails.push(format!("VNI   {vk:>10} -> {gv:>8}  want {want}"));
+        }
+    }
+    assert!(fails.is_empty(), "tone placement mismatches:\n{}", fails.join("\n"));
+}