How can the Raku behavior on capturing group in alternate be the same as Perl

user_17227456 0
Viewed 115

How can Raku behavior on capturing group in alternate be just like Perl regex' one e.g.

> 'abefo' ~~ /a [(b) | (c) (d)] (e)[(f)|(g)]/
｢abef｣
 0 => ｢b｣
 2 => ｢e｣
 3 => ｢f｣

needed to be 'usual' Perl regex result (let index system stay Raku):

 $0 = 'b'
 $1 = undef
 $2 = undef 
 $3 = e
 $4 = f

Please give useful guide.

raku
2 Answers

Quoting the Synopsis 5: Regexes and Rules design speculation document:

it is still possible to mimic the monotonic Perl 5 capture indexing semantics

Inserting a $3= for the (e):

/ a [ (b) | (c) (d) ] $3=(e) [ (f) | (g) ] /

andthen say 'abefo' ~~ $_

｢abef｣
 0 => ｢b｣
 3 => ｢e｣
 4 => ｢f｣

I've briefly looked for a mention of this in the doc but didn't see it.

So maybe we should file doc issues for mentioning this, presumably in Capture numbers and $ ($1, $2, ...).

user_10776720

Unclear as to the question, but surely going back to Perl5 semantics vs Raku means changing the alternation operator.

Perl5's | alternation operator is one in which the "first matching alternative" wins. The equivalent alternation operator in Raku is ||.

Raku's | alternation operator performs Longest Token Matching (LTM), (which roughly means if you separate your alternatives by |, you can spend less time ordering them by longest token to get the desired result).

https://docs.raku.org/language/regexes#Alternation:_||
https://docs.raku.org/language/regexes#Longest_alternation:_|

(As for capture numbering, maybe you can submit a request that that gets handled by the :Perl5 or :P5 regex adverb? See: https://docs.raku.org/language/regexes#Perl_compatibility_adverb )

user_72706490
Related