fix: trailing whitespaces are allowed for local_time (#3)

Input:

(each `·` indicates a whitespace)

```toml
valid·=·1979-05-27·
```

Output:

```
# before

(file
  (ERROR
    (key)))

# after

(file
  (pair
    (key) (local_date)))
```

.editorconfig

@@ -7,3 +7,6 @@ indent_size = 2
 indent_style = space
 insert_final_newline = true
 trim_trailing_whitespace = true
+
+[corpus/*]
+trim_trailing_whitespace = false

corpus/custom.txt

@@ -64,7 +64,7 @@ invalid2 = 1979-05-27 T 07:32:00 Z
 
 (file
   (ERROR
-    (key) (ERROR) (local_time) (local_time)
+    (key) (local_date) (local_time) (local_time)
   )
 )
 
@@ -78,11 +78,22 @@ invalid2 = 1979-05-27 T 07:32:00
 --------------------------------------------------------------------------------
 
 (file
-  (pair
+  (ERROR
-    (key) (ERROR) (local_time)
+    (key) (local_date) (local_time) (local_time)
   )
+)
+
+================================================================================
+VALID - local date - trailing whitespaces are allowed
+================================================================================
+
+valid = 1979-05-27 
+
+--------------------------------------------------------------------------------
+
+(file
   (pair
-    (key) (ERROR) (local_time)
+    (key) (local_date)
   )
 )
 

grammar.js

@@ -153,18 +153,14 @@ module.exports = grammar({
     boolean: $ => /true|false/,
 
     offset_date_time: $ =>
-      seq(
+      concatRegex(
         rfc3339_date,
-        token.immediate(rfc3339_delimiter),
+        rfc3339_delimiter,
-        token.immediate(rfc3339_time),
+        rfc3339_time,
-        token.immediate(rfc3339_offset)
+        rfc3339_offset
       ),
     local_date_time: $ =>
-      seq(
+      concatRegex(rfc3339_date, rfc3339_delimiter, rfc3339_time),
-        rfc3339_date,
-        token.immediate(rfc3339_delimiter),
-        token.immediate(rfc3339_time)
-      ),
     local_date: $ => rfc3339_date,
     local_time: $ => rfc3339_time,
 

src/grammar.json

@@ -563,57 +563,12 @@
       "value": "true|false"
     },
     "offset_date_time": {
-      "type": "SEQ",
+      "type": "PATTERN",
-      "members": [
+      "value": "(([0-9]+)-(0[1-9]|1[012])-(0[1-9]|[12][0-9]|3[01]))([ tT])(([01][0-9]|2[0-3]):([0-5][0-9]):([0-5][0-9]|60)([.][0-9]+)?)(([zZ])|([+-]([01][0-9]|2[0-3]):[0-5][0-9]))"
-        {
-          "type": "PATTERN",
-          "value": "([0-9]+)-(0[1-9]|1[012])-(0[1-9]|[12][0-9]|3[01])"
-        },
-        {
-          "type": "IMMEDIATE_TOKEN",
-          "content": {
-            "type": "PATTERN",
-            "value": "[ tT]"
-          }
-        },
-        {
-          "type": "IMMEDIATE_TOKEN",
-          "content": {
-            "type": "PATTERN",
-            "value": "([01][0-9]|2[0-3]):([0-5][0-9]):([0-5][0-9]|60)([.][0-9]+)?"
-          }
-        },
-        {
-          "type": "IMMEDIATE_TOKEN",
-          "content": {
-            "type": "PATTERN",
-            "value": "([zZ])|([+-]([01][0-9]|2[0-3]):[0-5][0-9])"
-          }
-        }
-      ]
     },
     "local_date_time": {
-      "type": "SEQ",
+      "type": "PATTERN",
-      "members": [
+      "value": "(([0-9]+)-(0[1-9]|1[012])-(0[1-9]|[12][0-9]|3[01]))([ tT])(([01][0-9]|2[0-3]):([0-5][0-9]):([0-5][0-9]|60)([.][0-9]+)?)"
-        {
-          "type": "PATTERN",
-          "value": "([0-9]+)-(0[1-9]|1[012])-(0[1-9]|[12][0-9]|3[01])"
-        },
-        {
-          "type": "IMMEDIATE_TOKEN",
-          "content": {
-            "type": "PATTERN",
-            "value": "[ tT]"
-          }
-        },
-        {
-          "type": "IMMEDIATE_TOKEN",
-          "content": {
-            "type": "PATTERN",
-            "value": "([01][0-9]|2[0-3]):([0-5][0-9]):([0-5][0-9]|60)([.][0-9]+)?"
-          }
-        }
-      ]
     },
     "local_date": {
       "type": "PATTERN",

src/node-types.json

@@ -132,21 +132,6 @@
       ]
     }
   },
-  {
-    "type": "local_date",
-    "named": true,
-    "fields": {}
-  },
-  {
-    "type": "local_date_time",
-    "named": true,
-    "fields": {}
-  },
-  {
-    "type": "offset_date_time",
-    "named": true,
-    "fields": {}
-  },
   {
     "type": "pair",
     "named": true,
@@ -327,6 +312,18 @@
     "type": "boolean",
     "named": true
   },
+  {
+    "type": "offset_date_time",
+    "named": true
+  },
+  {
+    "type": "local_date_time",
+    "named": true
+  },
+  {
+    "type": "local_date",
+    "named": true
+  },
   {
     "type": "local_time",
     "named": true

src/scanner.c

@@ -15,6 +15,10 @@ bool tree_sitter_toml_external_scanner_scan(
   TSLexer *lexer,
   const bool *valid_symbols
 ) {
+  while (lexer->lookahead == ' ' || lexer->lookahead == '\t') {
+    lexer->advance(lexer, true);
+  }
+
   if (lexer->lookahead == 0 || lexer->lookahead == '\n') {
     lexer->result_symbol = LINE_ENDING_OR_EOF;
     lexer->mark_end(lexer);