Tanenbaum

Tanenbaum is an OCaml starter project for Advent of Code.

• Automatic downloading + caching of puzzle inputs
• CLI for running puzzles + submitting answers to adventofcode.com
• Automatic puzzle detection + registration
• ReasonML support out of the box

First, clone or fork this repo and change directories into the repo.

$ opam switch create . --deps-only --y

$ opam install ocaml-lsp-server ocamlformat utop

$ dune build

Tanenbaum needs a session token from adventofcode.com in order to download your puzzle inputs and submit your answers.

Start by logging in to adventofcode.com, then browsing to one of the puzzle dashboards (e.g. https://adventofcode.com/2015).

Open your developer tools and reload the page. This should issue a GET request, e.g. GET https://adventofcode.com/2015. Look for the Cookie: request header, which should contain a value that looks like (where x is a hex value):

session=5xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx5

This value is your session token. You'll need to store it as an environment variable called AUTH_TOKEN. One convenient way of doing this is to use a tool like direnv, e.g.:

.envrc:

export AUTH_TOKEN="5xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx5"

$ mkdir inputs/{year}/{day}.txt

$ mkdir inputs/2023/01.txt

Each problem needs to conform to the Problem.T interface, which provides basic information about which year/day the problem is for, and functions to call which compute the puzzle output from puzzle input.

Tanenbaum will automatically register .ml or .re files in lib/problems which start with problem -- e.g. lib/problems/problem_2023_01.ml or lib/problems/problem_2023_01.re. Once you've added a file of this form, you can run it from the CLI.

For example, if we'd like to start working on day 1 of year 2023, Tanenbaum will automatically we can add this file:

lib/problems/problem_2023_01.ml;

let year = 2023
let day = 1

module Part_1 = struct
  let run (input : string) : (string, string) result = Ok input
end

module Part_2 = struct
  let run (input : string) : (string, string) result = Ok input
end

lib/problems/problem_2023_01.re;

let year = 2023;
let day = 1;

module Part_1 = {
  let run = (input: string): result(string, string) => Ok(input);
};

module Part_2 = {
  let run = (input: string): result(string, string) => Ok(input);
};

lib/problems/problem_2023_01.mli:

include Problem.T

lib/problems/problem_2023_01.rei:

include Problem.T;

Once you've added your problem, you can test your solution by running it with dune (optionally providing the --watch flag will re-run your problem when you change your code). This will output your answer to the terminal:

$ dune exec --watch bin/main.exe -- \
  --year=2023 \
  --day=1 \
  --part=1

  # output 
  your_answer

Once you're convinced that your problem produces the correct output, you can either copy and paste your solution into the answer form on adventofcode.com, or you can also submit your answer using the CLI via the --submit flag.

dune exec bin/main.exe -- \
  --year=2023 \
  --day=1 \
  --part=1 \
  --submit

Advent of Code typically provides smaller inputs, in order to check that your code works. I tend to allow Tanenbaum to download the puzzle input first, then I can replace the puzzle input with whatever input I'd like to test.

I can then revert back to the official puzzle input by deleting the file (Tanenbaum will download a fresh copy when I run it again).