Re^6: Why is "any" slow in this case?

[LanX]

> please test thoroughly, I just hacked the code into my mobile as an example ...

as expected the first approach was buggy.

This approach seems to pass the tests (using 1 for 0 to test more edge cases)

Hint: Anchoring is tricky when dealing with variable length strings and look-aheads.

use v5.14;
use warnings;

say "INPUT: ",my $str = join " ", 0 ..34;
my %DONT;
@DONT{1, 15, 16, 31}=();

my $stop=15;

my $donts = join "|", keys %DONT;
say my $dont_re = "(?!(?:$donts)\\b)";

  
while ( $str =~ m/\b$dont_re(\d+) \b$dont_re(\d+)/g ) {
  say "$1 $2";
  die unless $stop--;
  exists $DONT{$_} and die "$_ forbidden" for $1,$2;
}

INPUT: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 2
[phone] 32 33 34
(?!(?:31|1|15|16)\b)
2 3     # <-- no 0,1
4 5
6 7
8 9
10 11
12 13   # <-- no 14,15,16
17 18
19 20
21 22
23 24
25 26
27 28
29 30    # <-- no 31
32 33    # <-- no 34

UPDATE

had to improve the rexex again... :/

Probably a negative-look-behind is easier and faster.