osu-data 0.2.2

Containerize osu! data into a MySQL container and optionally serve .osu files via NGINX

osu! Data on Docker

pip install osu-data; osu-data -m mania -v top_1000 -ymd YYYY_MM_DD

Docker must be installed and running on your machine.

Retrieves database data from https://data.ppy.sh/ and hosts it on a local MySQL server. Optionally, store all ranked/loved .osu files in a service with the -f tag.

Get Started

IMPORTANT: MySQL data persists across runs. Recreate the MySQL Service if you changed the data used.

1. Install via pip pip install osu-data

2. Minimally, specify:

   • -m, --mode: The game mode to build the database with. osu, taiko, catch or mania
   • -v, --version: The database version. top_1000, top_10000 or random_10000

3. Optionally, specify:

   • -ymd, --year_month_day: The year, month, day of the database in the format YYYY_MM_DD
   • -p, --port: The port to expose MySQL on. Default is 3308
   • -f, --files: Whether to download .osu files.
   • -np, --nginx-port: The port to expose the nginx service on. Default is 8080. Not used if -f is not specified.
   • --...: See below table, these are optional flags to include or exclude more data. By specifying the flag, will INVERT the default value.

Option	Default Value
--beatmap-difficulty-attribs	False
--beatmap-difficulty	False
--scores	True
--beatmap-failtimes	False
--user-beatmap-playcount	False
--beatmaps	True
--beatmapsets	True
--user-stats	True
--sample-users	True
--counts	True
--difficulty-attribs	True
--beatmap-performance-blacklist	True

These options are chosen to be the most useful for analysis, and performance.

E.g.

osu-data \
  -m osu -v top_1000 -ymd 2023_08_01 -p 3308 -f \
  --beatmap-difficulty 

• Download the top 1000 osu! standard beatmaps
• from 1st August 2023
• expose MySQL on port 3308
• download .osu files
• include beatmap difficulty data

4. Connect on:
   • localhost:<MYSQL_PORT>
   • localhost:<NGINX_PORT> (if -f is specified)

Common Issues

• Docker daemon is not running. Make sure that Docker is installed and running. If you're using Docker Desktop, make sure it's actually started.
• MySQL Data isn't incorrect. A few reasons
  • Import was abruptly stopped. This can cause some .sql files to be missing / incomplete. Delete the whole compose project and try again.
  • Didn't specify the optional flags to include files. By default, some .sql files are not loaded. Take a look at osu-data -h and specify the optional flags to include them.
  • Data is outdated. By default, on every re-run of osu-data, the data is preserved. To update the data, you must delete the whole compose project and try again.
• wget: server returned error: HTTP/1.1 404 Not Found. This happens when you try to pull a YYYY_MM_DD that doesn't exist, and happens often when the data isn't yet ready on the start of each month. Check on https://data.ppy.sh/ to see which YYYY_MM_DD are available.
• rm: can't remove '../osu.mysql.init/*': This is safe to ignore.
• MySQL Credentials. By default, the MySQL doesn't have a password, so just use root as the username and leave the password blank.
• No files service. This is default, files service is optional and must be activated with the -f tag. osu-data -h for more info.

mysql.cnf

The database is tuned to be fast in importing speed, thus shouldn't be used for production. Notably, we set innodb_doublewrite = 0 which can compromise data integrity in the event of a crash. If you want to use this for production, we recommend to set this up from this Git repo, and tweak mysql.cnf.

Important Matters

1. Do not distribute the built images as per peppy's request. Instead, you can just share the code to build your image, which should yield the same result.
2. This database is meant to be for analysis, it's not tuned for production. Tweak mysql.cnf after importing for more MySQL customizations.
3. Finally, be mindful on conclusions you make from the data.

Changelog

• 0.1.5:
  • Allowed wider range of Python versions 3.9 ~ 4.0.
• 0.2.0:
  • Added GitHub Actions to automatically create dataset on workflow dispatch.
  • Year, Month specification is now Year, Month, Day because some data dumps don't fall exactly on day 1.
    • -ym -> -ymd, --year-month -> --year-month-day
    • Default of -ymd is removed to encourage users to check source of data.