Change Case follows each run's language for Turkic casing (#3479) (#3480)

Co-authored-by: aiongg <67543387+aiongg@users.noreply.github.com>

crates/engine/src/cmd/format.rs

@@ -7,6 +7,7 @@ use wordcraft_doc::resolve::ResolvedChar;
 use wordcraft_doc::{InlineObject, PartKind, Pos, RevisionKind, StoryRef};
 
 use super::sel_result;
+use super::tools::{is_turkic, lower_char, upper_char};
 use crate::{CmdError, CmdResult, CommandSpec, Session, p};
 
 /// Word's font size list.
@@ -453,7 +454,16 @@ fn change_case(s: &mut Session, v: &Value) -> CmdResult {
             let to = if path == b.path { b.off } else { p.len() };
             let chars = live_chars(p, from, to);
             let src: String = chars.iter().map(|(_, c)| c).collect();
-            case_groups(p, &chars, &convert_chars(&src, &mode, from == 0))
+            // Turkish, Azerbaijani and Crimean Tatar text follows the same case mapping as
+            // AutoCorrect (#1923); each run's own language decides (#3479).
+            let turkic: Vec<bool> = chars
+                .iter()
+                .map(|(off, _)| {
+                    let lang = s.doc.styles.resolve_char(p.props.style.as_deref(), p.props_of_char(*off)).lang;
+                    is_turkic(lang.as_deref().or(s.editing_lang.as_deref()))
+                })
+                .collect();
+            case_groups(p, &chars, &convert_chars(&src, &mode, from == 0, &turkic))
         };
         if groups.is_empty() {
             continue;
@@ -587,37 +597,47 @@ fn is_title(t: &str) -> bool {
 
 #[cfg(test)]
 fn convert(src: &str, mode: &str, para_start: bool) -> String {
-    convert_chars(src, mode, para_start).concat()
+    let turkic = vec![false; src.chars().count()];
+    convert_chars(src, mode, para_start, &turkic).concat()
 }
 
 /// The case-converted form of each character of `src`, in order (one entry per char, so a
-/// change of length such as ß → SS or İ → i̇ can be applied run by run).
-fn convert_chars(src: &str, mode: &str, para_start: bool) -> Vec<String> {
+/// change of length such as ß → SS or İ → i̇ can be applied run by run). `turkic` is one flag
+/// per char: true when that character's run is Turkish, Azerbaijani or Crimean Tatar.
+fn convert_chars(src: &str, mode: &str, para_start: bool, turkic: &[bool]) -> Vec<String> {
     let chars: Vec<char> = src.chars().collect();
     // Greek capital sigma lowers to the final form at the end of a word (as `str::to_lowercase`).
     let lower = |i: usize, c: char| -> String {
         let prev = i.checked_sub(1).and_then(|j| chars.get(j)).is_some_and(|p| p.is_alphabetic());
         let next = chars.get(i + 1).is_some_and(|n| n.is_alphabetic());
-        if c == 'Σ' && prev && !next { "ς".to_string() } else { c.to_lowercase().collect() }
+        if c == 'Σ' && prev && !next {
+            "ς".to_string()
+        } else if turkic.get(i).copied().unwrap_or(false) {
+            lower_char(c, true)
+        } else {
+            c.to_lowercase().collect()
+        }
     };
     // Title case capitalises the first word of any selection; sentence case only at a paragraph start.
     let mut up = mode == "title" || para_start;
     let mut out = Vec::with_capacity(chars.len());
     for (i, &c) in chars.iter().enumerate() {
+        let turkic = turkic.get(i).copied().unwrap_or(false);
+        let upper = |c: char| upper_char(c, turkic);
         let new = match mode {
-            "upper" => c.to_uppercase().collect(),
+            "upper" => upper(c),
             "lower" => lower(i, c),
             "toggle" => {
                 if c.is_uppercase() {
                     lower(i, c)
                 } else {
-                    c.to_uppercase().collect()
+                    upper(c)
                 }
             }
             "title" => {
                 let n = if up && c.is_alphabetic() {
                     up = false;
-                    c.to_uppercase().collect()
+                    upper(c)
                 } else {
                     lower(i, c)
                 };
@@ -630,7 +650,7 @@ fn convert_chars(src: &str, mode: &str, para_start: bool) -> Vec<String> {
                 // Sentence case.
                 let n = if up && c.is_alphabetic() {
                     up = false;
-                    c.to_uppercase().collect()
+                    upper(c)
                 } else {
                     lower(i, c)
                 };

crates/engine/src/cmd/tools.rs

@@ -1028,7 +1028,7 @@ fn prev_token(before: &str) -> String {
 }
 
 /// Turkish, Azerbaijani (and Crimean Tatar) map `i`/`İ` and `ı`/`I` as pairs (#1923).
-fn is_turkic(lang: Option<&str>) -> bool {
+pub(crate) fn is_turkic(lang: Option<&str>) -> bool {
     if wordcraft_proof::ProofLang::from_tag(lang) == wordcraft_proof::ProofLang::Tr {
         return true; // tr, tur, any region, same normalisation as proofing
     }
@@ -1038,7 +1038,7 @@ fn is_turkic(lang: Option<&str>) -> bool {
 }
 
 /// A letter in capitals: `i` → `İ` in Turkic languages, otherwise the Unicode default.
-fn upper_char(c: char, turkic: bool) -> String {
+pub(crate) fn upper_char(c: char, turkic: bool) -> String {
     if turkic && c == 'i' {
         return "\u{130}".to_string();
     }
@@ -1047,7 +1047,7 @@ fn upper_char(c: char, turkic: bool) -> String {
 
 /// A letter in lowercase: `I` → `ı` in Turkic languages. `İ` is always the single letter `i`,
 /// never `i` plus a combining dot.
-fn lower_char(c: char, turkic: bool) -> String {
+pub(crate) fn lower_char(c: char, turkic: bool) -> String {
     match c {
         'I' if turkic => "\u{131}".to_string(),
         '\u{130}' => "i".to_string(),

crates/engine/src/tests_typing.rs

@@ -2075,6 +2075,46 @@ fn autocorrect_capitalisation_follows_turkic_case_mapping() {
     assert_eq!(s.doc.plain_text(StoryRef::Body), "İngilizce ", "no combining dot in any language");
 }
 
+/// #3479: Format > Change Case follows the same Turkic case mapping as AutoCorrect, using
+/// each run's own language; other languages keep the Unicode default.
+#[test]
+fn change_case_follows_turkic_case_mapping() {
+    for tag in ["tr-TR", "az-AZ"] {
+        for (mode, input, want) in [
+            ("upper", "istanbul", "İSTANBUL"),
+            ("lower", "IŞIK", "ışık"),
+            ("title", "istanbul güzel", "İstanbul Güzel"),
+            ("sentence", "istanbul. izmir", "İstanbul. İzmir"),
+            ("toggle", "iI", "İı"),
+        ] {
+            let mut s = s();
+            run(&mut s, "review.language", json!({"lang": tag, "default": true}));
+            run(&mut s, "text.insert", json!({"text": input, "raw": true}));
+            run(&mut s, "select.all", json!({}));
+            run(&mut s, "format.changeCase", json!({"mode": mode}));
+            assert_eq!(s.doc.plain_text(StoryRef::Body), want, "{tag} {mode}");
+        }
+    }
+    // Other languages keep the Unicode default.
+    let mut s = self::s();
+    run(&mut s, "text.insert", json!({"text": "istanbul", "raw": true}));
+    run(&mut s, "select.all", json!({}));
+    run(&mut s, "format.changeCase", json!({"mode": "title"}));
+    assert_eq!(s.doc.plain_text(StoryRef::Body), "Istanbul", "en-US");
+    // Mixed runs: each run's own language decides.
+    let mut s = self::s();
+    run(&mut s, "text.insert", json!({"text": "IŞIK IŞIK", "raw": true}));
+    run(&mut s, "caret.set", json!({"pos": Pos::body(0, 0)}));
+    run(&mut s, "caret.set", json!({"pos": Pos::body(0, 5), "extend": true}));
+    run(&mut s, "review.language", json!({"lang": "tr-TR"}));
+    run(&mut s, "caret.set", json!({"pos": Pos::body(0, 6)}));
+    run(&mut s, "caret.set", json!({"pos": Pos::body(0, 11), "extend": true}));
+    run(&mut s, "review.language", json!({"lang": "en-US"}));
+    run(&mut s, "select.all", json!({}));
+    run(&mut s, "format.changeCase", json!({"mode": "lower"}));
+    assert_eq!(s.doc.plain_text(StoryRef::Body), "ışık işik", "mixed runs");
+}
+
 // Spelling as you type (#1552): the word being typed gets no red squiggle until the caret leaves
 // it; F7 and the Spelling pane still report it. See docs/typing-parity.md. Paragraphs start with
 // a correct word so AutoCorrect's sentence capital doesn't change the misspelled one.

docs/typing-parity.md

@@ -187,8 +187,9 @@ Applied when a word is finished by a space, punctuation **or Enter** (`enter_fin
   `İngilizce`, `ılık` → `Ilık`; observed in Word for Mac with Turkish proofing). Lowercasing `İ`
   gives the single letter `i` in every language, never `i` + combining dot. Other languages keep
   the Unicode default (`istanbul` → `Istanbul` in English)
-  (`autocorrect_capitalisation_follows_turkic_case_mapping`). Format > Change Case is not yet
-  Turkish-aware.
+  (`autocorrect_capitalisation_follows_turkic_case_mapping`); Format > Change Case follows the
+  same Turkic mapping from each run's language, while other languages keep the Unicode default
+  (`change_case_follows_turkic_case_mapping`, #3479).
 - Names of days are capitalised (`monday` → `Monday`). AutoCorrect Options (#412,
   `tools.autocorrect`, saved per user) switch each of these off: replacements, sentence and
   table-cell capitals, days, smart quotes, fractions, ordinals, dashes, links, bulleted and