fix: don't parse html comments inside regexes

tree-sitter · Feb 7, 2024 · aa186dc · aa186dc
1 parent 24b9885
commit aa186dc
Show file tree

Hide file tree

Showing 5 changed files with 28 additions and 10 deletions.
diff --git a/grammar.js b/grammar.js
@@ -19,9 +19,10 @@ module.exports = grammar({
     $._ternary_qmark,
     $.html_comment,
     '||',
-    // We use escape sequence to tell the scanner if we're currently inside a string or template string, in which case
+    // We use escape sequence and regex pattern to tell the scanner if we're currently inside a string or template string, in which case
     // it should NOT parse html comments.
     $.escape_sequence,
+    $.regex_pattern,
   ],
 
   extras: $ => [

diff --git a/src/grammar.json b/src/grammar.json
diff --git a/src/parser.c b/src/parser.c
diff --git a/src/scanner.c b/src/scanner.c
@@ -9,6 +9,7 @@ enum TokenType {
     HTML_COMMENT,
     LOGICAL_OR,
     ESCAPE_SEQUENCE,
+    REGEX_PATTERN,
 };
 
 void *tree_sitter_javascript_external_scanner_create() { return NULL; }
@@ -283,7 +284,8 @@ bool tree_sitter_javascript_external_scanner_scan(void *payload, TSLexer *lexer,
         return scan_ternary_qmark(lexer);
     }
 
-    if (valid_symbols[HTML_COMMENT] && !valid_symbols[LOGICAL_OR] && !valid_symbols[ESCAPE_SEQUENCE]) {
+    if (valid_symbols[HTML_COMMENT] && !valid_symbols[LOGICAL_OR] && !valid_symbols[ESCAPE_SEQUENCE] &&
+        !valid_symbols[REGEX_PATTERN]) {
         return scan_html_comment(lexer);
     }
 

diff --git a/test/corpus/statements.txt b/test/corpus/statements.txt
@@ -1022,6 +1022,7 @@ Comments within expressions
 
 y // comment
   * z;
+var a = /<!--/;
 
 ---
 
@@ -1030,7 +1031,12 @@ y // comment
     (binary_expression
       (identifier)
       (comment)
-      (identifier))))
+      (identifier)))
+  (variable_declaration
+    (variable_declarator
+      (identifier)
+      (regex
+        (regex_pattern)))))
 
 ==========================================
 HTML Comments