ohm-js: use Unicode letter properties in isSyntactic (#611)

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>

Author: pdubroy

doc/releases/ohm-js-18.0.md

@@ -147,6 +147,12 @@ if (result.failed()) {
 }
 ```
 
+## Breaking changes
+
+### Syntactic vs. lexical rule classification
+
+`isSyntactic()` now looks at the first _letter_ in the rule name — if it's an upper case letter, the rule is syntactic, otherwise it's lexical. In v17, we compared `firstChar === firstChar.toUpperCase()`, which incorrectly classified rule names starting with non-letter characters (e.g. `_`, digits) as syntactic. This now means that a rule named `_ident` is now treated as lexical, not syntactic.
+
 ## Removed APIs
 
 The following v17 APIs do not exist in v18 (yet):

packages/ohm-js/src/common.js

@@ -101,9 +101,13 @@ export function copyWithoutDuplicates(array) {
   return noDuplicates;
 }
 
+// Changed in v18 to use Unicode letter properties, matching the compiler.
+// Previously, this compared `firstChar === firstChar.toUpperCase()`, which
+// incorrectly returned true for non-letter characters (e.g. '_', digits).
 export function isSyntactic(ruleName) {
-  const firstChar = ruleName[0];
-  return firstChar === firstChar.toUpperCase();
+  const firstLetter = ruleName.match(/\p{L}/u)?.[0];
+  if (!firstLetter) return false;
+  return /\p{Lu}/u.test(firstLetter);
 }
 
 export function isLexical(ruleName) {