handle identifiers better

README.md

@@ -31,7 +31,8 @@ The aim is to build unit testcases from selected failure classes and slowly get
 
 ## Todo
 
-* [ ] use [Unicode® Standard Annex #31](https://www.unicode.org/reports/tr31/) (augmented with '-') for identifiers
+* [x] use [Unicode® Standard Annex #31](https://www.unicode.org/reports/tr31/) (augmented with '-') for identifiers
+  * allow starting '_' also, not to spec but terraform allows it
 * [ ] add [template expressions](https://github.com/hashicorp/hcl/blob/main/hclsyntax/spec.md#template-expressions)
   * [x] add quoted templates
     * [x] add quoted template interpolations

grammar.js

@@ -34,14 +34,13 @@ module.exports = grammar({
   ],
 
   rules: {
-    config_file: $ => optional($.body),
+    // also allow objects to handle .tfvars in json format
+    config_file: $ => optional(choice($.body, $.object)),
 
     body: $ => repeat1(
       choice(
         $.attribute,
         $.block,
-        // not to spec but handles .tfvars in json format
-        $.object,
       ),
     ),
 
@@ -62,10 +61,9 @@ module.exports = grammar({
     _block_start: $ => '{',
     _block_end: $ => '}',
 
-    // TODO: not to spec but good enough for now
     identifier: $ => token(seq(
-      choice(/\p{L}/, '_'),
-      repeat(choice(/\p{L}/, /[0-9]/, /(-|_)/)),
+      choice(/\p{ID_Start}/, '_'),
+      repeat(choice(/\p{ID_Continue}/, '-')),
     )),
 
     expression: $ => prec.right(choice(

src/grammar.json

@@ -5,8 +5,17 @@
       "type": "CHOICE",
       "members": [
         {
-          "type": "SYMBOL",
-          "name": "body"
+          "type": "CHOICE",
+          "members": [
+            {
+              "type": "SYMBOL",
+              "name": "body"
+            },
+            {
+              "type": "SYMBOL",
+              "name": "object"
+            }
+          ]
         },
         {
           "type": "BLANK"
@@ -25,10 +34,6 @@
           {
             "type": "SYMBOL",
             "name": "block"
-          },
-          {
-            "type": "SYMBOL",
-            "name": "object"
           }
         ]
       }
@@ -113,7 +118,7 @@
             "members": [
               {
                 "type": "PATTERN",
-                "value": "\\p{L}"
+                "value": "\\p{ID_Start}"
               },
               {
                 "type": "STRING",
@@ -128,15 +133,11 @@
               "members": [
                 {
                   "type": "PATTERN",
-                  "value": "\\p{L}"
+                  "value": "\\p{ID_Continue}"
                 },
                 {
-                  "type": "PATTERN",
-                  "value": "[0-9]"
-                },
-                {
-                  "type": "PATTERN",
-                  "value": "(-|_)"
+                  "type": "STRING",
+                  "value": "-"
                 }
               ]
             }

src/node-types.json

@@ -130,10 +130,6 @@
         {
           "type": "block",
           "named": true
-        },
-        {
-          "type": "object",
-          "named": true
         }
       ]
     }
@@ -188,6 +184,10 @@
         {
           "type": "body",
           "named": true
+        },
+        {
+          "type": "object",
+          "named": true
         }
       ]
     }