in reply to Find Length Of Longest Ascending/Descending Sequence

Not a working solution, just a thought experiment:

Perl 6 has a regex construct spelled <*abcd>. It is a shortcut for a|ab|abc|abcd (also note that in Perl 6 regexes, the longest pipe-delimited alternative wins, not the first one as in Perl 5).

Sadly no compiler implements the <*abcd> feature yet, so I can't demonstrate it, but I think something like this should work: 

my token longest_asc_desc {
    # ascending
    | <*0123456789> ** <?after 9>
    | <*1234567890> ** <?after 0>
    | <*2345678901> ** <?after 1>
    ...
    # descending:
    | <*9876543210> ** <?after 0>
    | <*8765432109> ** <?after 9>
    | <*7654321098> ** <?after 8>
    ...
}

[download]
Perl 6 - second systems done right

Replies are listed 'Best First'.
Re^2: Find Length Of Longest Ascending/Descending Sequence
by Ratazong (Monsignor) on May 10, 2011 at 09:22 UTC

    nice feature :-)

    How will it match data like the following? 

    11101234567890123411111   =>  expected result: 15

    [download]
    Will you need to create the pattern in <*0123456789> dynamically based on the length of the data?

    Rata
[reply]
[d/l]
[select]
      Notice the repetition: 
      <*0123456789> ** <?after 9>

      [download]

      is short for 

      <*0123456789> [ <?after 9> <*0123456789>]*

      [download]

      So if it matches the first one, and upto 9 (that's what the <?after 9> checks), it tries again to start with 0.

      To extract the longest match, you need to collect all, and overlapping (because the last digits of an ascending sequence can be the first of a descending sequence): 

      my regex seq {
   ...
}
my $length = [max] $teststring.match(&seq, :g, :overlap)>>.chars;

      [download]
      Perl 6 - second systems done right
[reply]
[d/l]
[select]

In Section Seekers of Perl Wisdom