advoco 0.3.0

Advent of Code helper library. There are many like it, but this one is mine.

Advoco

Advent of Code helper library - why not?? Dear future contributor traipsing through history, this was when this was a private repo that me and a buddy were collaborating on. If this README does not look more polished in the current version, hunt me down and slap me.

Install

poetry install

I'm still working actively so do this often.

tl;dr

Do this once a year:

advoco generate-year

Then grab your session cookie from AoC using browser tools and add it to the environment as AOC_TOKEN. Put it in your .bashrc and never think about it again (until next year). The session lasts like a month so you shouldn't have to do this that often.

Then, every day, from the directory created with the generate-year command:

advoco generate -d <day you want to generate>

or if it's after midnight, AoC time (Eastern):

advoco generate

In the generated script file, you'll find a function for part1 and part2 of the daily AoC puzzle. You'll also find a transform function. This function will act on each line of the input. For instance, input always comes in as a string, so if you want your puzzle input to be a list of ints, do this:

def transform(line):
    return int(line)

The last thing you'll find in the script is the call to advoco.do(). It starts out with sample=True and will therefore attempt to parse the sample input Eric uses on the webpage of the daily puzzle. It'll throw an error and give you advice if it can't find sample input. When you're ready to try your solutions on your real input, remove the sample argument.

Explanations

Some context...

WTF is do doing?

When you call do(), advoco inspects the call stack for the file that called it. It looks through that file (as a module, using getattr) for a few things:

• A transform func. If it doesn't find one, the default one is basically lambda x: return x
• Some part functions, i.e. part1, part2

It gets the raw input for the puzzle you're working on (more on that below), runs that through whatever transform it's using, then passes it as the arg to the part functions it found in your script file.

Finally, it gets the result of your part function and prints it. It does not yet attempt to submit the answer for you, and it may not do that ever (not sure yet).

WHAT YEAR (AND DAY) IS IT???

Advoco needs to figure out which puzzle input to go get, both the puzzle year and the puzzle day. To determine this - really in all instances where advoco needs to figure out a year and/or day - the following order is honored:

1. Anything explicitly passed. If you pass a day or year arg to do, that is what advoco is going to work with. End of discussion. Explicit declaration trumps all.
2. The calling context. Where is do being called from? Where is a cli command being invoked? If do is being called from a python file, and that file name contains digits, those digits are assumed to represent the day you want to work on. Further, if that python file lives in a directory whose name contains the regex r"20\d{2}", then that regex match is assumed to be the year. For cli invocations, the folder == year convention holds, but the day is never inferred from calling context.
3. The current day and year, in the AoC timezone (America/New_York).

Notes

This still assumes that we're gonna be working with a list of lines as our input. I haven't yet figured out the best way to handle the case where you want the whole damn input string as one blob. Good ideas are welcome here.

Any input fetched from the AoC site is cached in ~/.cache/advoco to not pester the site. The inputs, as far as I can tell, have never and will never change, so there's no cache expiry mechanism.

The generate-year command creates a top level folder called _localonly_2022. This is just so you can AoC from within the advoco repo without git stressing out. When it's ready for prime time, I'll change it to make a 2022 directory.

Contributing

Bug reports are awesome but if you see something and you know how to fix it, by all means! Just make a new branch and open a PR. Hell, just push directly to main if you're super confident.