Advent of Code

All Years - Blazingly Fast

Runtimes

History

When I first started Advent in 2020 I wrote everything in Python for simplicity as well as getting better at my main scripting language.

I then decided to go back and solve all the years from 2015 on also in Python.

Since then I have used Advent as an opportunity to learn new languages.

So far this has been:

• 2021: Go
• 2022: Rust
• 2023: OCaml
• 2024: Zig

Sometime after 2022 I decided to go back and optimize every solution with the goal of having no single solution take more than 1 second to run. Sometimes this involves algorithm improvements, other times I decided to re-write solutions from Python into something like Rust or Go. Currently the slowest solution takes just under 750 ms, so mission accomplished.

Input Data

The input for each puzzle is in a separate private Github repo

This is done to comply with the policy here.

These inputs end up in the data folder in the same structure as the solutions.

For instance the input for year 2020 day 5 is file data/2020/05/data.txt.

After cloning this repo the following command needs to be run to get the data.

git submodule update --init

This also means when pulling changes the update command must also be ran.

git submodule update

Setup

Install Requirements

pip install -r scripts/requirements.txt

Set Aliases

alias a-build="./scripts/advent.py build"
alias a-run="./scripts/advent.py run"
alias a-gen="./scripts/advent.py generate"
alias a-graph="./scripts/advent.py graph"

Build

The build target is used to do any language specific setup prior to running.

This includes downloading any necessary libraries, compiling targets, etc.

None of the parameters are required, the default behavior in this case is to build and test all supported languages.

• Alias Command: a-build
• Direct Command: ./scripts/advent.py build

Usage

a-build \
  (--language <language>)* \
  --info?

Variable Name	Alt	Description	Default	Example
language	-l	Limit build to the specified languages	None	-l rust
info	-i	Outputs which languages will be built	False	-i

Run

The run target is used to run various days rather than running directly.

Does some hacky stuff to set arguments and run commands, but for the most part runs standard build commands for each language / framework.

None of the parameters are required, the default behavior in this case is to run the latest day in the fastest language it is implemented in.

• Alias Command: a-run
• Direct Command: ./scripts/advent.py run

Usage

a-run \
  --template <template>? \
  (--year <year>)* \
  (--day <day>)* \
  (--language <language>)* \
  --strategy <strategy>? \
  --slow <time>? \
  --test? \
  --info?

Variable Name	Alt	Description	Default	Example
template	-t	Name that targets specific years / days	latest	-t languages
year	-y	List of years to run	None	-y 2021 -y 2022
day	-d	List of days to run	None	-d 1 -d 3 -d 5
language	-l	Limit runs to the specified languages	None	-l go
strategy	-s	A way to select which languages to run	None	-s fastest
slow	-S	Defines the runtime (in ms) for slow	100	-S 500
test	-T	Passes test flag to each day	False	-T
info	-i	Outputs which days would run	False	-i

• If template is provided then year & day must not be provided
• If year or day are provided then template must not be provided

Process Runtime Output

cat all.json | jq -r '.[]|[.year, .day, .language, .runtime, .execution]|@tsv'
cat all.json | jq -r '.[]|[.year, .day, .language, .runtime]|@tsv' | sort -nk4
cat all.json | jq -r '.[]|[.year, .day, .language, .runtime]|@tsv' | sort -nk4 | awk '{ if ($4 > 100) { print $0 } }'
cat all.json | jq -r '.[]|select(.year == 2015 and .day == 24)'
cat all.json | jq -r '.[]|.runtime' | awk '{ sum+=$1 } END { print sum, "ms" }'
cat all.json | jq -r '.[]|select(.year == 2015)|.runtime' | awk '{ sum+=$1 } END { print sum, "ms" }'
cat all.json | jq -r '.[]|[.year, .runtime]|@tsv' | awk '{ data[$1] += $2 } END { for (year in data) { print year, data[year], "ms" } }'

Unit Test

This will test that some shared logic works across usage days. Such as the int-code implementation from 2019.

pytest -s scripts

Generate

The generate target creates initial files and empty data file for the specified language.

Will do any other required setup, such as updating Cargo.toml for rust.

Will pull down your puzzle input if instructions are followed.

None of the parameters are required, the default behavior in this case is to generate the next day using the rust template.

• Alias Command: a-gen
• Direct Command: ./scripts/advent.py generate

Usage

a-gen \
  --template <template>? \
  --year <year>? \
  --day <day>? \
  --language <language>? \
  --puzzle? \
  --info?

Variable Name	Alt	Description	Default	Example
template	-t	Name that targets specific year / day	next	-t next
year	-y	Year to generate starting files for	None	-y 2022
day	-d	Day to generate starting files for	None	-d 5
language	-l	Language to generate starting files for	rust	-l python
puzzle	-p	Download puzzle description as well	False	-p
info	-i	Outputs which day would get generated	False	-i

• If template is provided then year & day must not be provided
• If year or day are provided then template must not be provided

Install aoc-cli

Template generation script can use aoc-cli to download input: docs.

Follow the instructions in the README of aoc-cli to get your session cookie setup.

This library relies on openssl which you will also need to install if you don't already have it.

cargo install aoc-cli
touch .adventofcode.session

Graph

The graph target creates a variety of graphs to visualize the runtime of days split and grouped on different dimensions.

None of the parameters are required, the default behavior in this case is to create only new graphs and skip graphs that have been created before.

• Alias Command: a-graph
• Direct Command: ./scripts/advent.py graph

Usage

a-graph \
  --archive? \
  --info?

Variable Name	Alt	Description	Default	Example
archive	-a	Archive existing graphs	False	-a
info	-i	Outputs whether graphs would be arhived	False	-i