Course of Raku / Regexes and grammars / Regexes / Captures

Non-capturing groups

Round brackets ( ) do two jobs at once: they group a part of the pattern and they capture it into $0, $1, and so on. Sometimes you only want the first job — to treat several atoms as a single unit — without spending a capture number on it. For that, use square brackets [ ], which group without capturing.

Grouping matters as soon as you attach a quantifier. A quantifier binds only to the atom right before it, so the pattern below repeats just the final letter:

say 'abccc' ~~ / abc ** 3 /; # ｢abccc｣

Only the c was repeated three times. Wrap the whole piece in [ ] to repeat all of it:

say 'abcabcabc' ~~ / [ abc ] ** 3 /; # ｢abcabcabc｣

Now [ abc ] is one unit, and ** 3 applies to the group.

Because [ ] takes no capture slot, the numbers stay reserved for the parts you actually care about. Compare a plain key–value match:

if 'foo=42' ~~ / (\w+) '=' (\d+) / {
    say $0; # ｢foo｣
    say $1; # ｢42｣
}

Suppose the key may be preceded by a word you want to skip. Group that prefix with [ ] so it does not disturb the numbering:

if 'the foo=42' ~~ / [ \w+ \s ]? (\w+) '=' (\d+) / {
    say $0; # ｢foo｣
    say $1; # ｢42｣
}

The optional [ \w+ \s ]? matched the but claimed no capture, so $0 is still the key and $1 is still the value. Had you written that prefix with ( ), everything would have shifted: the prefix would become $0, the key $1, and the value $2.

The rule is short: use ( ) when you want to keep the matched text, and [ ] when you only want to group. Both kinds accept quantifiers and alternations, which is why the earlier alternation [ cat | dog ] house used square brackets — it needed the grouping but not another capture.

Course navigation

← Quiz — Captures   |   Split a filename →

💪 Or jump directly to the exercises in this section.