fix: parse sublist with empty line without error (resolve #23)

Author: emiasims

corpus/basic.tst

@@ -814,6 +814,27 @@ List.8b  - Whitespace after text
         (paragraph
           (expr))))))
 
+================================================================================
+List.9  - newline before sub listitem
+================================================================================
+- a
+
+  - b
+--------------------------------------------------------------------------------
+
+(document
+  (body
+    (list
+      (listitem
+        (bullet)
+        (paragraph
+          (expr))
+        (list
+          (listitem
+            (bullet)
+            (paragraph
+              (expr))))))))
+
 ================================================================================
 Directive.1  - Document
 ================================================================================

grammar.js

@@ -241,14 +241,6 @@ org_grammar = {
       seq($.listitem, $._listend)
     ),
 
-    listitem: $ => seq(
-      $.bullet,
-      choice(
-        $._eof,
-        alias($.body, "item_body"),
-      )
-    ),
-
     listitem: $ => seq(
       field('bullet', $.bullet),
       choice(

src/scanner.cc

@@ -1,9 +1,6 @@
 #include <tree_sitter/parser.h>
 #include <vector>
 #include <cwctype>
-#include <cstring>
-#include <cassert>
-#include <stdio.h>
 
 namespace {
 
@@ -180,8 +177,8 @@ bool scan(TSLexer *lexer, const bool *valid_symbols) {
   // 1. dedent
   // 2. same indent, not a bullet
   // 3. two eols
+  int16_t newlines = 0;
   if (valid_symbols[LISTEND] || valid_symbols[LISTITEMEND]) {
-    int16_t newlines = 0;
     for (;;) {
       if (lexer->lookahead == ' ') {
         indent_length++;
@@ -232,8 +229,7 @@ bool scan(TSLexer *lexer, const bool *valid_symbols) {
   }
 
   // - Liststart and bullets
-  if (valid_symbols[LISTSTART] || valid_symbols[BULLET]) {
-
+  if ((valid_symbols[LISTSTART] || valid_symbols[BULLET]) && newlines == 0) {
     Bullet bullet = getbullet(lexer);
 
     if (valid_symbols[BULLET] && bullet == bullet_stack.back() && indent_length == indent_length_stack.back()) {