Escape issue for characeters

[rogerfar]

My parser contains the following characters:

WS                      : (' '|'\r'|'\t'|'\u000C'|'\n') -> skip;
SQ                      : ['];
DQ                      : '\u0022';
SP                      : '\u0020';
HTAB                    : '\u0009';
CR                      : '\u000D';
LF                      : '\u000A';

When I target C# the code is good and compiles fine, no errors in C# or Antlr.

But when I compile it against Java, I get:

error: unclosed character literal "'false'", "'null'", null, null, "'\u0022'", "'\u0020'", "'\u0009'",

When I look in the generated Java file I see:

private static String[] makeLiteralNames() {
	return new String[] {
		null, null, null, "'datetimeoffset'", "'datetime'", "'guid'", "'true'", 
		"'false'", "'null'", null, null, "'\u0022'", "'\u0020'", "'\u0009'", 
		"'\u000D'", "'\u000A'", null, null, null, "'$'", null, null, null, null, 
		"'('", "')'", "'['", "']'", "'{'", "'}'", "'~'", null, null, "'/'", "'.'", 
		"':'", "'%'", "'@'", "'!'", "'?'", "'_'", null, null, null, null, null, 
		null, "'asc'", "'desc'", "'and'", "'or'", "'eq'", "'ne'", "'lt'", "'le'", 
		"'gt'", "'ge'", "'in'", "'nin'", "'nlike'", "'like'"
	};
}

The issue is that "'\u0022'" is not valid Java, this should be '\u0022'.

Now this can be mitigated by changing the grammer to:

WS                      : (' '|'\r'|'\t'|'\u000C'|'\n') -> skip;
SQ                      : ['];
DQ                      : [\u0022];
SP                      : [\u0020];
HTAB                    : [\u0009];
CR                      : [\u000D];
LF                      : [\u000A];

But is this the recommended approach or a bug?

[KvanTTT]

It looks duplicate to #2281, it's a bug.

The issue is that "'\u0022'" is not valid Java, this should be '\u0022'.

No it should be "'\\u0022'". Your literal also throws compiler erros.

[KvanTTT]

@parrt also close