If you

1. sort your search patterns by length and then value (reversed);
2. chop characters off the string one at a time until it is less than the longest/highest string;
3. then descend down the array comparing against each string in turn;
4. quit the loop when you find a match, or loop and repeat;

The worst case time I've seen is always less than 2 milliseconds. How does that compare with the regex solution?

#! perl -slw
use strict;
use Math::Random::MT qw[ rand ];
use Benchmark::Timer;

my @array = sort{
    length( $b ) <=> length( $a )
    ||
    $b cmp $a
} qw[ 0123456 012345 01234], map int rand( 123456789 ), 1 .. 5022;

my $str = '012345666666';

my $T = new Benchmark::Timer;
$T->start( '11-char string in 5025 possibles' );
my $copy = $str;
my $n = 0;
while( $n < $#array and $copy ne $array[ $n ] ) {
    chop $copy while $copy gt $array[ $n ];
    $n++ while $copy lt $array[ $n ];
    last if $copy eq $array[ $n ];
}
if( length $copy ) {
    print "Best match found $copy (at posn: $n )";
}
else {
    print "No match found";
}
$T->stop( '11-char string in 5025 possibles' );
$T->report;

print for grep defined, @array[ $n - 10 .. $n + 10 ];

__END__
C:\test>583928
Best match found 0123456 (at posn: 4975 )
1 trial of 11-char string in 5025 possibles (1.621ms total)

1166599
1157595
1146996
1065416
1063127
1063055
1052870
1034509
1021521
1004863
0123456
997625
973306
962972
943968
911619
881691
873333
838373
832949
788294
[download]