Grammars

Grammars organise regexes into named rules — a real parser as a language feature.

A grammar is a class whose methods are regexes. TOP is the entry point, and .parse matches the whole string against it:

grammar Numbers {
    token TOP { <num>+ % [',' \s*] }
    token num { \d+ }
}

my $m = Numbers.parse('1, 22, 333');
say so $m;
say $m<num>.map(+*).sum;

True
356

<num> calls the num token by name; + % [',' \s*] means "one or more, separated by commas". Every named rule that matched is available on the match object — $m<num> is the list of all three numbers.

A tiny config parser #

Nested rules build real structure. Here's an entire key–value file format:

grammar Config {
    token TOP   { <entry>+ }
    token entry { <key> '=' <value> \n? }
    token key   { \w+ }
    token value { <-[\n]>+ }
}

my $conf = Config.parse("name=Camelia\ncolor=purple\n");
for $conf<entry>.list -> $e {
    say "{ $e<key> } is { $e<value> }";
}

name is Camelia
color is purple

<-[\n]> is a negated character class — "anything but a newline". Grammars are how Raku programs parse configuration files, log lines, and — in more than one real project — entire programming languages.

Try it #

Extend Sums so it also allows - between numbers (square brackets group alternatives: ['+' || '-']), then make the test string parse.

grammar Sums {
    token TOP { <num>+ % '+' }
    token num { \d+ }
}

say so Sums.parse('1+2-3');

grammar Sums {
    token TOP { <num>+ % ['+' || '-'] }
    token num { \d+ }
}

say so Sums.parse('1+2-3');

True