Handle strings starting with whitespaces - antlr

Question

I'm trying to create an ANTLR v4 grammar with the following set of rules:
1.In case a line starts with #, it is considered a label:
#label
2.In case the line starts with cmd, it is treated as a command
cmd param1 param2
3.If a line starts with a whitespace, it is considered a string. All the text should be extracted. Strings can be multiline, so they end with an empty line
A long string with multiline support
and any special characters one can imagine.
<-empty line here->
4.Lastly, in case a line starts with anything but whitespace, # and cmd, it's first word should be considered a heading.
Heading A long string with multiline support
and any special characters one can imagine.
<-empty line here->
It was easy to handle lables and commands. But I am clueless about strings and headings.
What is the best way to separate whitespace word whitespace whatever doubleNewline and whatever doubleNewline? I've seen a lot of samples with whitespaces, but none of them works with both random text and newlines. I don't expect you to write actual code for me. Suggesting an approach will do.

Answer 1

Something like this should do the trick:
lexer grammar DemoLexer;
LABEL
: '#' [a-zA-Z]+
;
CMD
: 'cmd' ~[\r\n]+
;
STRING
: ' ' .*? NL NL
;
HEADING
: ( ~[# \t\r\nc] | 'c' ~'m' | 'cm' ~'d' ).*? NL NL
;
SPACE
: [ \t\r\n] -> skip
;
OTHER
: .
;
fragment NL
: '\r'? '\n'
| '\r'
;
This does not mandate the "beginning of the line" requirement. If that is something you want, you'll have to add semantic predicates to your grammar, which ties it to a target language. For Java, that would look like this:
LABEL
: {getCharPositionInLine() == 0}? '#' [a-zA-Z]+
;
See:
Semantic predicates in ANTLR4?
https://github.com/antlr/antlr4/blob/master/doc/predicates.md