In part 2 I could only think of replacing the jokers with all unique values in the hand

Heh, better than I did, I just brute forced and tried every card possible

/shrug

I used DataFrames with Explorer for the first time today. May be overkill, but I learned a lot.

It was nice to be able to tinker directly in the results via LiveBook:

example1200×857 50.7 KB

Today’s puzzle was fun!

My solution:

I accidentally removed my code, so I wrote it again and didn’t bother to refactor it.

Prep

{:ok, puzzle_input} =
  KinoAOC.download_puzzle("2023", "7", System.fetch_env!("LB_AOC_SESSION"))

strengths =
  ?2..?9
  |> Map.new(&{&1, &1 - ?0})
  |> Map.merge(%{
    ?T => 10,
    ?J => 11,
    ?Q => 12,
    ?K => 13,
    ?A => 14
  })

Part 1

puzzle_input
|> String.split()
|> Stream.chunk_every(2)
|> Stream.map(fn [hand, bid] ->
  cards = String.to_charlist(hand)
  freqs = cards |> Enum.frequencies() |> Map.values() |> Enum.sort(:desc)
  scores = Enum.map(cards, &strengths[&1])
  {freqs, scores, String.to_integer(bid)}
end)
|> Enum.sort()
|> Stream.map(&elem(&1, 2))
|> Stream.with_index(1)
|> Stream.map(fn {bid, rank} -> bid * rank end)
|> Enum.sum()

Part 2

strengths = %{strengths | ?J => 1}

puzzle_input
|> String.split()
|> Stream.chunk_every(2)
|> Stream.map(fn [hand, bid] ->
  cards = String.to_charlist(hand)
  freqs = cards |> Enum.frequencies()
  {jokers, freqs} = Map.pop(freqs, ?J, 0)
  freqs = freqs |> Map.values() |> Enum.sort(:desc)
  freqs = (freqs == []) && [5] || [hd(freqs) + jokers | tl(freqs)]
  scores = Enum.map(cards, &strengths[&1])
  {freqs, scores, String.to_integer(bid)}
end)
|> Enum.sort()
|> Stream.map(&elem(&1, 2))
|> Stream.with_index(1)
|> Stream.map(fn {bid, rank} -> bid * rank end)
|> Enum.sum()

I loved doing this. The way I decided to calculate rank through pattern matching ended up being visual cue to compute the J assisted ranking.

advent_of_code/lib/2023/day_07.ex at master · code-shoily/advent_of_code (github.com)

This felt so natural!

One valuable lesson learned though.

My smaller? function had @card_rank_1 as default parameter. And it is recursive, so when I called it with @card_rank_2 during part 2, I forgot to provide the second parameter on the recursive calls, so subsequent matches were lies! And to make it worse, sample data were immune to this ranking algorithm and kept giving me right answer so I spent some time in frustration.