fix(editor): include fullwidth parens in completion depth

apps/desktop/src/lib/sql/insertValueHints.ts

@@ -486,13 +486,14 @@ function stepLexState(sql: string, index: number, state: LexState, dialectId: st
     }
     return index + 1;
   }
-  if (ch === "(") {
+  // Match tokenizeSqlSemantic: fullwidth （） update nesting the same as ASCII ().
+  if (ch === "(" || ch === "\uFF08") {
     state.depth += 1;
-    return index + 1;
+    return index + ch.length;
   }
-  if (ch === ")") {
+  if (ch === ")" || ch === "\uFF09") {
     state.depth = Math.max(0, state.depth - 1);
-    return index + 1;
+    return index + ch.length;
   }
   return index + 1;
 }

apps/desktop/src/lib/sql/semantic/tokens.ts

@@ -232,11 +232,14 @@ export function tokenizeSqlSemantic(input: string, dialectId = "mysql", options?
       continue;
     }
 
-    if ("(),.;*".includes(ch)) {
-      if (ch === ")") depth = Math.max(0, depth - 1);
-      tokens.push(token("punctuation", ch, start, start + 1, depth));
-      if (ch === "(") depth += 1;
-      index += 1;
+    // Fullwidth （） (U+FF08/U+FF09) are common IME typos in Chinese locales; treat them like
+    // ASCII () for nesting depth so commas inside to_date（…, …） stay nested and do not split
+    // INSERT ... SELECT projections (insert value column hints).
+    if ("(),.;*".includes(ch) || ch === "\uFF08" || ch === "\uFF09") {
+      if (ch === ")" || ch === "\uFF09") depth = Math.max(0, depth - 1);
+      tokens.push(token("punctuation", ch, start, start + ch.length, depth));
+      if (ch === "(" || ch === "\uFF08") depth += 1;
+      index += ch.length;
       continue;
     }
 

packages/app-tests/insertValueHints.test.ts

@@ -279,6 +279,42 @@ test("keeps the hint for a projection whose alias names a different column", ()
   );
 });
 
+test("does not split nested to_date arguments into extra SELECT projections", () => {
+  // ASCII parentheses: commas inside to_date(to_char(...), 'fmt') stay nested.
+  const ascii = [
+    "INSERT INTO t_demo (occur_date, company_no, company_name, dep_id, market_code)",
+    "SELECT",
+    "  to_date(to_char(a.l_date), 'YYYYMMDD') as occur_date,",
+    "  '01694' as company_no,",
+    "  'Acme Corp' as company_name,",
+    "  d.dept_code as dep_id,",
+    "  a.market_code",
+    "FROM dual a, dual d",
+  ].join("\n");
+  assert.deepEqual(
+    parseInsertValueHints(ascii).map((hint) => ({ column: hint.column, text: ascii.slice(hint.from).split(/[\s,]/u, 1)[0] })),
+    [{ column: "market_code", text: "a.market_code" }],
+  );
+
+  // Fullwidth outer parentheses (common IME typo) must not treat the format-string comma as a
+  // top-level projection separator — that shifted every later hint by one.
+  const fullwidth = [
+    "INSERT INTO t_demo (occur_date, company_no, company_name, dep_id, market_code)",
+    "SELECT",
+    "  to_date\uFF08to_char(a.l_date), 'YYYYMMDD'\uFF09 as occur_date,",
+    "  '01694' as company_no,",
+    "  'Acme Corp' as company_name,",
+    "  d.dept_code as dep_id,",
+    "  a.market_code",
+    "FROM dual a, dual d",
+  ].join("\n");
+  assert.equal(parseInsertValuesClauses(fullwidth)[0]?.rows[0]?.length, 5);
+  assert.deepEqual(
+    parseInsertValueHints(fullwidth).map((hint) => ({ column: hint.column, text: fullwidth.slice(hint.from).split(/[\s,]/u, 1)[0] })),
+    [{ column: "market_code", text: "a.market_code" }],
+  );
+});
+
 test("caps INSERT ... SELECT hints to the smaller target or projection count", () => {
   assert.deepEqual(
     parseInsertValueHints("INSERT INTO t (a, b) SELECT x, y, z FROM source").map((hint) => hint.column),