Add u|U|u8 string|char support (#36)

grammar.js

@@ -773,26 +773,26 @@ module.exports = grammar({
     },
 
     char_literal: $ => seq(
-      choice('L\'', '\''),
+      choice('L\'', 'u\'', 'U\'', 'u8\'', '\''),
       choice(
         $.escape_sequence,
         token.immediate(/[^\n']/)
       ),
       '\''
     ),
-
+    
     concatenated_string: $ => seq(
       $.string_literal,
       repeat1($.string_literal)
     ),
 
     string_literal: $ => seq(
-      choice('L"', '"'),
+      choice('L"', 'u"', 'U"', 'u8"', '"'),
       repeat(choice(
         token.immediate(prec(1, /[^\\"\n]+/)),
         $.escape_sequence
       )),
-      '"'
+      '"',
     ),
 
     escape_sequence: $ => token.immediate(seq(

src/grammar.json

@@ -4905,6 +4905,18 @@
               "type": "STRING",
               "value": "L'"
             },
+            {
+              "type": "STRING",
+              "value": "u'"
+            },
+            {
+              "type": "STRING",
+              "value": "U'"
+            },
+            {
+              "type": "STRING",
+              "value": "u8'"
+            },
             {
               "type": "STRING",
               "value": "'"
@@ -4959,6 +4971,18 @@
               "type": "STRING",
               "value": "L\""
             },
+            {
+              "type": "STRING",
+              "value": "u\""
+            },
+            {
+              "type": "STRING",
+              "value": "U\""
+            },
+            {
+              "type": "STRING",
+              "value": "u8\""
+            },
             {
               "type": "STRING",
               "value": "\""

src/node-types.json

@@ -2893,6 +2893,14 @@
     "type": "L'",
     "named": false
   },
+  {
+    "type": "U\"",
+    "named": false
+  },
+  {
+    "type": "U'",
+    "named": false
+  },
   {
     "type": "[",
     "named": false
@@ -3069,6 +3077,22 @@
     "type": "typedef",
     "named": false
   },
+  {
+    "type": "u\"",
+    "named": false
+  },
+  {
+    "type": "u'",
+    "named": false
+  },
+  {
+    "type": "u8\"",
+    "named": false
+  },
+  {
+    "type": "u8'",
+    "named": false
+  },
   {
     "type": "union",
     "named": false

test/corpus/expressions.txt

@@ -118,6 +118,9 @@ int main() {
   "b" "c" "d";
   "\"hi\"";
   L"bonjour";
+  u"guten morgen";
+  U"buenos dias";
+  u8"buongiorno";
 }
 
 ---
@@ -130,6 +133,9 @@ int main() {
       (expression_statement (string_literal))
       (expression_statement (concatenated_string (string_literal) (string_literal) (string_literal)))
       (expression_statement (string_literal (escape_sequence) (escape_sequence)))
+      (expression_statement (string_literal))
+      (expression_statement (string_literal))
+      (expression_statement (string_literal))
       (expression_statement (string_literal)))))
 
 ============================================
@@ -142,6 +148,9 @@ int main() {
   '\t';
   '\'';
   L'b';
+  u'c';
+  U'\xa1';
+  u8'\x1A';
 }
 
 ---
@@ -155,7 +164,10 @@ int main() {
       (expression_statement (char_literal (escape_sequence)))
       (expression_statement (char_literal (escape_sequence)))
       (expression_statement (char_literal (escape_sequence)))
-      (expression_statement (char_literal)))))
+      (expression_statement (char_literal))
+      (expression_statement (char_literal))
+      (expression_statement (char_literal (escape_sequence)))
+      (expression_statement (char_literal (escape_sequence))))))
 
 ============================================
 Field access