Re: the case where regex seems to work slower

[tybalt89]

Just fudge things a little :)

#!/usr/bin/perl

# http://perlmonks.org/?node_id=1196089

use strict;
use warnings;

$\ = $/;

my $dict = 'b';
$_ = '*' . 'a' x (9e4 - 1e4) . "\n";

my($head, $tail) = map length( $_ // '' ), split /\*/, $_;

s/\?/[$dict]/g;
s/\*/ [^$dict ]* /;

my $qr = qr/^$_$/;

substr($_, $head, 0, ' '),
  substr($_, -$tail, 0, ' '),
  print $_ =~ $qr ? "YES" : "NO"
    for 'a' x (0 + 1e4) . 'a' x (9e4 - 1e4) . "\n";

This avoids the massive backtrack this particular problem encounters.

[rsFalse]

I think this do not work at cases when '*' is absent or not in the middle!
Read more... (4 kB)

[tybalt89]

Hey, it worked for the test case you gave :)

(grumble, grumble, inadequate test cases :( Even these test cases are inadequate, can you figure out why? )

#!/usr/bin/perl

# http://perlmonks.org/?node_id=1196089

use strict;
use warnings;

$\ = $/;

while(<DATA>)
  {
  chomp;
  my $dict = $_;

  $_ = <DATA>;

  my($head, $tail) = map length( $_ // '' ), split /\*/;

  s/\?/[$dict]/g;
  s/\*/ [^$dict ]* /;

  my $qr = qr/^$_/;

  $_ = <DATA>,
    $head <= length($_) && substr($_, $head, 0, ' '),
    $tail && $tail <= length($_) && substr($_, -$tail, 0, ' '),
    print $_ =~ $qr ? "YES" : "NO"
      for 1 .. <DATA>;
  }

__DATA__
ab
a?a
3
aaa
aab
aaac
ab
*a?a
3
aaa
aab
caaa
ab
a?a*
3
aaa
aab
aaac
b
aaaaaaaaaaaaaaaaaaaaaaa*aaaaaaaaaaaaaaaaaa
1
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
b
aaaaaaaaaaaaaaaaaaaaaaa*aaaaaaaaaaaaaaaaaa
1
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
b
*aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
b
*aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
b
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa*
1
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
b
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa*
1
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
b
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
b
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa

[rsFalse]

Sorry, I didn't understand your second sentence, because parentheses were not balanced ;)

Thank you for improvement to cope with other cases. I've submitted, and it took only 171 ms at worst TC, this is #93 - http://codeforces.com/contest/832/submission/28930515

Upd.: In this solution there single spaces in regex are used. They were mysterious for me in previous code, and should be taken by negate character class.