Re^7: A regex that only matches at offset that are multiples of a given N? (Update:almost perfect!)

[smls]

If it was possible to avoid running 3 orders of magnitude slower

Well, in this benchmark the embedded code gets called on every single character in the string, while in the problem at hand it would only get called once for every correct match. Finding the match in the first place would happen completely within the regex engine.

If you are using the regex with a while loop, you could move the pos()-incrementing into the already existing loop body to avoid creating an additional scope:

while( $input =~ m{\G(?:.{$n})*?(?=fred(....))}g ) {
    pos($input) += $n;

    # ...
    # do stuff with $1
    # ...
}

This could also be adapted to ensure that new occurrences of "fred" may not be matched inside the (....) that belongs to the previous match, by modifying the pos()-incrementing line like this:

    pos($input) += $n * int((length('fred'.$1) - 1) / $n + 1);

[BrowserUk]

You're right!

Moving the pos modification into the while loop body means that it only gets executed after successful (aligned) matches.

And that helps my performance a lot.

Thank you. Especially for persisting when I wasn't seeing the wood for the trees.