Merge pull request #127 from lf-/jade/unicode-ids

fix: implement Unicode identifiers

grammar.js

@@ -982,7 +982,7 @@ module.exports = grammar({
     false: $ => token(choice('FALSE', 'false')),
     null: $ => 'NULL',
 
-    identifier: $ => /[a-zA-Z_]\w*/,
+    identifier: $ => /(\p{XID_Start}|_)\p{XID_Continue}*/,
 
     _type_identifier: $ => alias($.identifier, $.type_identifier),
     _field_identifier: $ => alias($.identifier, $.field_identifier),

src/grammar.json

@@ -6425,7 +6425,7 @@
     },
     "identifier": {
       "type": "PATTERN",
-      "value": "[a-zA-Z_]\\w*"
+      "value": "(\\p{XID_Start}|_)\\p{XID_Continue}*"
     },
     "_type_identifier": {
       "type": "ALIAS",

test/corpus/expressions.txt

@@ -52,6 +52,25 @@ int main() {
 
 ---
 
+(translation_unit
+  (function_definition
+    (primitive_type)
+    (function_declarator (identifier) (parameter_list))
+    (compound_statement
+      (expression_statement (identifier))
+      (expression_statement (identifier)))))
+
+============================================
+Unicode Identifiers
+============================================
+
+int main() {
+  µs;
+  blah_accenté;
+}
+
+---
+
 (translation_unit
   (function_definition
     (primitive_type)