AoC 2021/3 - Aiming for reuse


On with Advent of Code puzzle 3 from 2021: aiming for reuse of code.

This day’s puzzle introduces another theme that we will encounter again during this year’s installment: the wall of text. There must be some trick that the fast people use to sift through all that text, but I honestly don’t know.

#!/usr/bin/env raku
use v6;

sub MAIN ($filename = $?FILE.subst(/\.raku$/, '.sample')) {
   my $inputs = get-inputs($filename);
   my ($part1, $part2) = solve($inputs);

   my $highlight = "\e[1;97;45m";
   my $reset     = "\e[0m";
   put "part1 $highlight$part1$reset";
   put "part2 $highlight$part2$reset";

sub get-inputs ($filename) {
   $filename.IO.basename.IO.lines».comb(/<[0 1]>/)».Array;
} ## end sub get_inputs ($filename = undef)

sub solve ($inputs) {
   return (part1($inputs), part2($inputs));

sub part1 ($inputs) {
   my @benchmarks = $inputs.elems / 2 xx $inputs[0].elems;
   my @sums = [Z+] @$inputs;
   my $epsilon = ((@sums Z< @benchmarks)».Int).join('');
   my $gamma = TR/01/10/ given $epsilon;
   return $epsilon.parse-base(2) * $gamma.parse-base(2);

sub part2 ($inputs) {
   my $result = 1;
   for 0, 1 -> $t {
      my @candidates = @$inputs;
      for 0 .. @candidates[0].end -> $bit {
         my @s; # array of arrays of arrays, top indexed by 0, 1
         @s[$_[$bit]].push: $_ for @candidates;
         @candidates = @(@s[0] <= @s[1] ?? @s[$t] !! @s[1 - $t]);
         if (@candidates.elems == 1) {
            $result *= @candidates[0].join('').parse-base(2);
            next TARGET;
   return $result;

The first part is about considering each bit from the inputs, in isolation to the other bits. To figure out whether there are more 0 or 1 values, it’s sufficient to sum them all and compare against the half of their number. This explains the @benchmarks.

There’s a lot of showing off in this part 1, e.g. the hyper-application of the zipped version of the sum. A compact way to sum all the inputs bit by bit (assuming, as it is in this case, that all input sequences are stored as arrays of 0 and 1 values).

Calculating $epsilon is easy by comparing each bit position with the benchmark. Again, we leverage the Zip operator here, though not in its hyper form.

It’s interesting to note that $gamma is the bitwise complement of $epsilon, so this is how we calculate it.

Part 2 is a bit more… convoluted and I could not figure out how to show-off a few tricks. So there we go, with traditionally nested for loops as well as if conditions etc. etc. At each stage going ahead, we work on a different “vertical” slice like before, this time restricted to the the “survivors” in the previous pass.

So… this completes the 2021 edition of Advent of Code. Did you enjoy it? Complete it? Get crazy for it? Let me know!

Comments? Octodon, Twitter, GitHub, Reddit, or drop me a line!