[xpg934]

A comment that directly follows a string literal is not removed and breaks parsing:

var a = 'x'// comment          // SyntaxError: Unexpected token
var a = 'x'/* comment */       // SyntaxError: Unexpected token
var phone = '7900'//contact.phone ? contact.phone.replace(/\D/g, '') : 0;
                               // ArgumentException: Invalid flag in regexp definition

The same code with a space or a semicolon before the comment ('x' // comment, 'x';// comment) works, as does a comment right after a number or an identifier (1// comment). Reproduced on 2.5.1674, 2.5.1699 and 2.6.1722.

Cause

Parser.RemoveComments calls ValidateString, which on success already moves index past the closing quote, and then unconditionally does index++. So the first character after a string literal is skipped. If it is the / of a comment, SkipComment no longer sees the comment, the scan stops at the next / (taken for a regexp), and the parser later treats the comment as a division or a regexp literal.

Fix

Advance the index only when no string was consumed.

Tests

Added ParserTests.CommentRightAfterStringLiteral (6 cases, all failing before the fix). Full suite on net10.0 with the T4-generated tests: before — 7 failed (the 6 new ones and ShouldProduceCorrectResult), after — only ShouldProduceCorrectResult, which fails the same way without this change.

[nilproject]

Thanks!