fix: comment optimization should not cause automatic-semi insert for single-line block comment

Co-authored-by: Amaan Qureshi <[email protected]>

Author: jackschu

src/scanner.c

@@ -47,7 +47,18 @@ static bool scan_template_chars(TSLexer *lexer) {
     }
 }
 
-static bool scan_whitespace_and_comments(TSLexer *lexer, bool *scanned_comment) {
+typedef enum {
+    REJECT,     // Semicolon is illegal, ie a syntax error occurred
+    NO_NEWLINE, // Unclear if semicolon will be legal, continue
+    ACCEPT,     // Semicolon is legal, assuming a comment was encountered
+} WhitespaceResult;
+
+/**
+ * @param consume If false, only consume enough to check if comment indicates semicolon-legality
+ */
+static WhitespaceResult scan_whitespace_and_comments(TSLexer *lexer, bool *scanned_comment, bool consume) {
+    bool saw_block_newline = false;
+
     for (;;) {
         while (iswspace(lexer->lookahead)) {
             skip(lexer);
@@ -71,17 +82,25 @@ static bool scan_whitespace_and_comments(TSLexer *lexer, bool *scanned_comment)
                         if (lexer->lookahead == '/') {
                             skip(lexer);
                             *scanned_comment = true;
+
+                            if (lexer->lookahead != '/' && !consume) {
+                                return saw_block_newline ? ACCEPT : NO_NEWLINE;
+                            }
+
                             break;
                         }
+                    } else if (lexer->lookahead == '\n' || lexer->lookahead == 0x2028 || lexer->lookahead == 0x2029) {
+                        saw_block_newline = true;
+                        skip(lexer);
                     } else {
                         skip(lexer);
                     }
                 }
             } else {
-                return false;
+                return REJECT;
             }
         } else {
-            return true;
+            return ACCEPT;
         }
     }
 }
@@ -96,10 +115,12 @@ static bool scan_automatic_semicolon(TSLexer *lexer, bool comment_condition, boo
         }
 
         if (lexer->lookahead == '/') {
-            if (!scan_whitespace_and_comments(lexer, scanned_comment)) {
+            WhitespaceResult result = scan_whitespace_and_comments(lexer, scanned_comment, false);
+            if (result == REJECT) {
                 return false;
             }
-            if (comment_condition && lexer->lookahead != ',' && lexer->lookahead != '=') {
+
+            if (result == ACCEPT && comment_condition && lexer->lookahead != ',' && lexer->lookahead != '=') {
                 return true;
             }
         }
@@ -125,7 +146,7 @@ static bool scan_automatic_semicolon(TSLexer *lexer, bool comment_condition, boo
 
     skip(lexer);
 
-    if (!scan_whitespace_and_comments(lexer, scanned_comment)) {
+    if (scan_whitespace_and_comments(lexer, scanned_comment, true) == REJECT) {
         return false;
     }
 

test/corpus/semicolon_insertion.txt

@@ -272,6 +272,16 @@ let b /* comment between declarators */, c
 
 let d
 
+let e
+/* back to back *//* comments */
+
+class C {
+  method/*comment*/() {}
+}
+
+b
+/* interleaved non-semi-insertion */
+  .c
 ---
 
 (program
@@ -286,4 +296,14 @@ let d
   (comment)
   (comment)
   (comment)
-  (lexical_declaration (variable_declarator (identifier))))
+  (lexical_declaration (variable_declarator (identifier)))
+  (lexical_declaration (variable_declarator (identifier)))
+  (comment)
+  (comment)
+  (class_declaration (identifier) (class_body (method_definition
+    (property_identifier)
+    (comment)
+    (formal_parameters)
+    (statement_block))))
+  (expression_statement
+    (member_expression (identifier) (comment) (property_identifier))))