http://www.perlmonks.org?node_id=1018660


in reply to A regex that only matches at offset that are multiples of a given N?

Alright, here is a revised version of johngg's solution, that prevents the redundant double matching, while at the same time keeping the matching logic self-contained within the regex.

Instead of moving pos() forward manually (like I suggested in the discussion thread for johngg's solution), it lets the regex engine do this implicitly by having it gobble up $n characters (if available) after matching the zero-width look-ahead that contains the capture group:

# 0 5 10 15 20 25 30 35 # ' ' ' ' ' ' ' ' $_ = q{.....fred1..fred2...fred3....fred4..}; # ----++++----||||----||||----++++---- $n = 4 # -----||||+-----+++++||||-+++++-----+ $n = 5 my $capture = qr([0-9]\.+); # the (....) in the OP's specification for my $n ( 4, 5 ) { say "\$n = $n"; while ( m[\G(?:.{$n})*?(?=fred($capture)).{0,$n}]g ) { say " matched 'fred$1' at pos @{[pos($_)-$n]} (gobbled '$&')"; } }

Output:

$n = 4 matched 'fred2...' at pos 12 (gobbled '.....fred1..fred') matched 'fred3....' at pos 20 (gobbled '2...fred') $n = 5 matched 'fred1..' at pos 5 (gobbled '.....fred1') matched 'fred3....' at pos 20 (gobbled '..fred2...fred3')

Note that if length("fred$1") > $n, it will actually start looking for the next "fred" while still whithin the part matched by $1. If this must be avoided, I guess manual pos()-incrementing is still the best bet.