TL;DR

Here we are with TASK #1 from The Weekly Challenge #177. Enjoy!

The challenge

You are given a positive number, $n. Write a script to validate the given number against the included check digit. Please checkout the wikipedia page for information. Example 1 Input:$n = 5724
Output: 1 as it is valid number


Example 2

Input: $n = 5727 Output: 0 as it is invalid number  The questions If this were an interview question, I’d probably ask something about what we’re supposed to validate and what would go beyond these boundaries. The solution The Damm Algorithm is a way to calculate a control digit out of other digits. There are quite a few of these procedures, and it seems that this particular one has its own advantages, except - maybe - its reluctance on accepting leading 0, i.e. treating 1 and 0001 as if they’re the same thing. Which, honestly, they are - at least as long as we consider the input a positive number. In this challenge, it is. OK, this said, we have to play a treasure hunt onto a matrix. In Raku: #!/usr/bin/env raku use v6; sub MAIN ($input = 5724) { put damm-calculate($input) ?? 0 !! 1 } sub damm-calculate ($input) {
state @qs = <
0  3  1  7  5  9  8  6  4  2
7  0  9  2  1  5  4  8  6  3
4  2  0  6  8  7  1  3  5  9
1  7  5  0  9  8  3  4  2  6
6  1  2  3  0  4  5  9  7  8
3  6  7  4  2  0  9  5  8  1
5  8  6  9  7  2  0  1  3  4
8  9  4  5  3  6  2  0  1  7
9  4  3  8  6  1  7  2  0  5
2  5  8  1  4  3  6  7  9  0
>;
(0, $input.comb.Slip).reduce({@qs[10 *$^a + $^b]}); }  I love reduce, as it’s perfect for this kind of chain reaction stuff. We have to inject the equivalent of a leading 0 to get the ball started properly and focus on the first row at the beginning, but apart from this it’s just the usual take result and new item, rinse, repeat. The matrix is represented as a linear array, as I like doing lately. It’s just for having some fun, I think some time ago I saw there is no real benefit in doing this. It does not improve readability, anyway, but might help keeping attention levels high with a little what the hell moment. I have to admit that I slapped the Slip without too much thinking. I don’t know if I’m finally getting the gist of it, but I’ve been biten by the tightness of Raku containers many times and it’s a sort of defensive attempt from my side. Punishment-based learning, if you will, although I’d like to use Slip and flat knowing better what they are for. I’ll get there, one day. The Perl counterpart is pretty much a translation with functions instead of methods. It kind of flows more naturally, although backwards, because it does not force the construction of a container of digits to ignite the process. This is such a subtle pet peeve that I have troubles understanding it, too. #!/usr/bin/env perl use v5.24; use warnings; use experimental 'signatures'; no warnings 'experimental::signatures'; use List::Util 'reduce'; my$input = shift // 5724;
say damm_calculate($input) ? 0 : 1; sub damm_calculate ($input) {
state $qs = [qw< 0 3 1 7 5 9 8 6 4 2 7 0 9 2 1 5 4 8 6 3 4 2 0 6 8 7 1 3 5 9 1 7 5 0 9 8 3 4 2 6 6 1 2 3 0 4 5 9 7 8 3 6 7 4 2 0 9 5 8 1 5 8 6 9 7 2 0 1 3 4 8 9 4 5 3 6 2 0 1 7 9 4 3 8 6 1 7 2 0 5 2 5 8 1 4 3 6 7 9 0 >]; reduce {$qs->[10 * $a +$b]} 0, split m{}mxs, \$input;
}


And that’s all, folks!