twitter-text-parser.pl

use Data::Dumper;
my @r;

my $text = q{@hoge http://example.com/#@hoge/ aaa #hash aaa @hogehoge aaa http://example.com/};
my $re = qr{(http://[A-Za-z0-9_\-\~\.\%\?\#\@/]+(?{push @r, ["HTTP", @-,@+] })|(?:^|\s)\@[A-Za-z0-9_]+(?{push @r,["AT", @-,@+]})|(?:^|\s)#[A-Za-z0-9_]+(?{push @r,["HashTag",@-,@+]}))};

warn $text;
() = $text =~/$re/g;
warn Dumper @r;

for (@r) {
    warn substr $text, $_->[1], ($_->[2] - $_->[1]);
}

__END__

@hoge http://example.com/#@hoge/ aaa #hash aaa @hogehoge aaa http://example.com/ at twitter-text-parser.pl line 7.
$VAR1 = [
          'AT',
          '0',
          '5',
          undef
        ];
$VAR2 = [
          'HTTP',
          '6',
          '32',
          undef
        ];
$VAR3 = [
          'HashTag',
          '36',
          '42',
          undef
        ];
$VAR4 = [
          'AT',
          '46',
          '56',
          undef
        ];
$VAR5 = [
          'HTTP',
          '61',
          '80',
          undef
        ];
@hoge at twitter-text-parser.pl line 12.
http://example.com/#@hoge/ at twitter-text-parser.pl line 12.
 #hash at twitter-text-parser.pl line 12.
 @hogehoge at twitter-text-parser.pl line 12.
http://example.com/ at twitter-text-parser.pl line 12.