The yearly repo, seeded with a few solvers for random problems from prior years. If you don't know what Advent of Code is, you should go see! It's both lovely, and will help make sense of this repo. :)
This repo's current solve status:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 │ #
──────┬────────────────────────────────────────────────────────────────────────────┼─────
2015 │ * . . . . * .| . . . . . . . . * . . . . . . . . . │ 6
2016 │ * . . . . . . . . . * * *| . . . . . * . . . . . . │ 10
2017 │ * * .| . . . . . . . . . . . . . . . . . . . * . . │ 6
2018 │ . . . . * . . . . * .| . . . . . * . . . . . . . . │ 6
2019 │ . * . . . . . * .| . . . . * . . . . . ? . . . . . │ 6
2020 │ * . . * . . . . . . * .| . . . . . * . . . * . . . │ 10
2021 │ * * .| . . . * . . . . . . . . . . . . . . . . . . │ 6
2022 │ * . . . . . . . * .| . . * . . . . . . * . . . . . │ 8
2023 │ * . * . . . . . . . . .| . . * . . . . . . . * . . │ 8
2024 │ * * * * * * * * * *| * * * * * * * * * * │ 40
──────┼────────────────────────────────────────────────────────────────────────────┼─────
│ 16 8 4 4 4 4 4 4 4 4 6| 4 6 4 4 4 4 4 4 4 . 2 4 . . │ 106
Next? Donut Maze (https://adventofcode.com/2019/day/20)
https://github.com/barneyb/aoc2017 has an index of all years' repositories.
The main solvers are implemented as Rust modules, one per day. They leverage
advent-of-code-data, a Python
library for interacting with the adventofcode.com
API. Retrieving input comes out of the box; there's a small wrapper script src
that helps with and submitting answers. This alleviates a lot of manual steps,
as well as keeps all of my "private" info out of the source code.
The general idea is to run ./new_day.py, which sets up a day to work on, get
it solved, then run ./done.py. The generated skeleton will take care of both
submission and re-validation of your input, along with whatever examples are
available.
The ./new_day.py script accepts year and day params if you want to set up a
specific puzzle. With no options, it'll select the latest unsolved day from this
year, if one exists. Otherwise, one that is "far away" from what you've already
solved will be chosen. 2019, in particular, has dependencies between puzzles;
you won't get later puzzles unless you've solved the earlier ones.
You need a current-ish Rust, a current-ish Python, and aocd both pip-installed
and configured with your AoC token.
While there are binary solvers, a given solver module's tests are where to start.
When your token expires in ~30 days, you'll get completely non-handled 400 HTTP errors. Go update your token.
If you're especially masochistic, you can configure aocd with multiple account
tokens and run them all across every solver, all at once. Be careful, however,
as this may cause the server to rate-limit your user(s). Judicious use of
Ctrl-C is advised, until you have inputs and answers cached locally.
% ./run_all.py
Building..................................................Done! 138.77 ms
| github | gmail
2024 ==========================================================
2 Red-Nosed Reports | ✔ | ✔ 2.06 ms
1 Historian Hysteria | ✔ | ✔ 668.54 µs
2015 ==========================================================
1 Not Quite Lisp | ✔ | ✔ 251.92 µs
===============================================================
Success! 3 days x 2 accounts = 12 stars! 2.54 sec
Note that "fast" solvers' times are appreciably inflated by reporting overhead. On my machine, running Not Quite Lisp takes ~15µs to solve (vs the ~125µs suggested above), plus another ~30µs of thread overhead:
% cargo run -r --bin not_quite_lisp --quiet
Part A: 280 ( 11.583µs)
Part B: 1797 ( 3.417µs)
Exit ( 43.333µs)
Some of the binaries do visualization, rather than spit out answers. Run as
above, and hit s to capture your own screenshot to the viz folder. Don't
forget the -r in there; rustc's optimizations are both quick and impressive.
One example is probably_a_fire_hazard (2015 day 6),
which displays the final winning light pattern. Here's mine:
