Re^2: Find what characters never appear

[Narveson]

If we've seen $chr once, can we somehow avoid repeating the assignment to $seen[ord($chr)] during the rest of the read?

Can we avoid even testing $seen[ord($chr)]?

I'd like to make a regex that matches any of our dwindling array of unseen characters, and update this regex every time I update $seen. Has anybody done this?

[kennethk]

If you want to avoid potential issues w/ regex metacharacters, you can use a set of hash keys to track what's been seen and rebuild the regex once for each character:

#!/usr/bin/perl 
use strict;
use warnings;

my %char_hash = ();
$char_hash{ chr($_) } = undef foreach (33 .. 127);
my $chars = join "", keys %char_hash;
my $regex = "([\Q$chars\E])";

while (<DATA>) {
    while (/$regex/g) {
        delete $char_hash{$1};
        $chars = join "", keys %char_hash;
        $regex = "([\Q$chars\E])";
    }
}
my @good_array = keys %char_hash;
print @good_array;

__DATA__
!"#$%&'()*+,-./01234567
89:;<=>?@ABCDE
FGHIJKLMOPQRSTUVWXYZ[\]^_`abcdefghijklmnop
qrstuvwxyz{|}~

though I feel like there must be a simpler way of implementing this approach.

[Narveson]

This ran in just a few minutes against my big 2GB file.

All I had to do was change the printable range to 33..126, change <DATA> to <>, and for my own curiosity, add print "$1 seen on line $.\n"; after delete $char_hash{$1};

[almut]

Maybe something like this  (demo with reduced charset):

#!/usr/bin/perl

my $s   = "fccccaaaaeaaaddaaaaabbcccaaacaaabbaaaa";
my $set = "[abcdefg]";

while ($s =~ /($set)/g) {
    my $ch = $1;
    $set =~ s/$ch//;  # remove $ch from search set
    printf "found %s at %d -> regex now: %s\n", $ch, pos($s), $set;
}

__END__
found f at 1 -> regex now: [abcdeg]
found c at 2 -> regex now: [abdeg]
found a at 6 -> regex now: [bdeg]
found e at 10 -> regex now: [bdg]
found d at 14 -> regex now: [bg]
found b at 21 -> regex now: [g]

Update: kennethk noted that you would run into complications with regex metacharacters with this simple approach (when using the full ASCII set) — which is of course correct...