rosu-memory

Lightweight, cross-platform and simple osu! memory reader written in Rust

Usage

Currently rosu-memory tries to auto-detect your osu! folder from running osu! process. This means you can simply run it without providing additional arguments

But just in case you can always overwrite auto-detected path. Instructions on how to do that are listed below.

On Linux auto-detection doesn't work in cutting edge

How do I use counters/overlays?

Select any ported counter/overlay (you can see some inside examples folder)
Open OBS
Add to your scene Browser Source
Click on Local File checkbox
Choose index.html file inside counter/overlay folder
Done!

Setting path to the osu! folder manually

Windows

There are two primary methods

By setting environment value in windows settings (recommended):
1. Type Edit the system environment variables in your search bar, press enter
2. In opened window click Environment variables at the very bottom
3. Click New... under System Variables
4. In Variable name type OSU_PATH
5. In Variable value type your path to osu! folder
6. Click Ok
7. Now you can start rosu-memory simply by double clicking rosu-memory.exe file!
By editing batch file:
1. Open windows_start.bat file in your preferred text editor
2. Change set OSU_PATH=%HOMEDRIVE%%HOMEPATH%\AppData\Local\osu! to set OSU_PATH=<YOUR OSU PATH>
3. Start rosu-memory by opening windows_start.bat file (Note: rosu-memory.exe and windows_start.bat should be in the same folder)
  Linux
Download binary or compile it by yourself
Set OSU_PATH environment variable or run rosu-memory with --osu_path argument

Differences between gosumemory and tosu

Small binary size

rosu-memory gosumemory (v 1.3.8) tosu (v 1.3.0)

Windows 0.9 MB 21 MB 35 MB

Linux 1.2 MB 21.2 MB Does not support
Low RAM usage. Since Rust is a compiled language and doesn't have any garbage collection (as opposed to gosu and tosu), it has very low (almost minimal) RAM usage.
Low CPU usage
It's blazingly fast compared to other readers. We are using gradual pp calculation, which means we don't recalculate the whole beatmap each iteration, only calculating new objects that have been passed.
Background image appears almost instantly (if you used other readers you know that this is really big problem for them)
Statically linked binary, no runtime dependencies required (except win32 and glibc)
Can be compiled using musl libc!
Cross-platform
Different JSON schema. Small disadvantage, but rosu-memory outputs slightly different JSON than tosu and gosu. Please consider this when porting your counter to rosu-memory. (See examples folder :) )

	rosu-memory	gosumemory (v 1.3.8)	tosu (v 1.3.0)
Windows	0.9 MB	21 MB	35 MB
Linux	1.2 MB	21.2 MB	Does not support

What is not implemented yet

Not all osu! data is present yet. But everything that is required to make simple counter is present :)
Strains graph
Any tournament stuff
And many others :)

JSON Schema

{
    "skin": "lain's skin",
    "playtime": 78624,
    "menu_mode": 0,
    "state": 5,
    "stars": 6.772433558139333,
    "stars_mods": 6.772433558139333,
    "current_stars": 6.486382070844266,
    "result_screen": {
        "username": "",
        "mods": 0,
        "mode": 0,
        "max_combo": 0,
        "score": 0,
        "hit_300": 0,
        "hit_100": 0,
        "hit_50": 0,
        "hit_geki": 0,
        "hit_katu": 0,
        "hit_miss": 0,
        "accuracy": 0
    },
    "gameplay": {
        "mods": 0,
        "username": "Guest",
        "score": 0,
        "hit_300": 0,
        "hit_100": 0,
        "hit_50": 0,
        "hit_geki": 0,
        "hit_katu": 0,
        "hit_miss": 0,
        "accuracy": 0.9141824751580849,
        "combo": 0,
        "max_combo": 0,
        "mode": 0,
        "slider_breaks": 0,
        "unstable_rate": 0,
        "passed_objects": 0,
        "grade": "B",
        "current_hp": 0,
        "current_hp_smooth": 0
    },
    "beatmap": {
        "artist": "Eternal Tears Of Sorrow",
        "title": "Another Me",
        "creator": "LMT",
        "difficulty": "Insane 1.36x (250bpm) CS4.2 AR9.6 OD9",
        "map_id": 2643167,
        "mapset_id": 1271974,
        "ar": 9.6,
        "cs": 4.2,
        "hp": 5,
        "od": 9,
        "beatmap_status": 2,
        "last_obj_time": 167310,
        "first_obj_time": 150,
        "bpm": 250,
        "paths": {
            "beatmap_full_path": "/path/to/osu/Songs/1271974 Eternal Tears Of Sorrow - Another Me/Eternal Tears Of Sorrow - Another Me (LMT) [Insane 1.36x (250bpm) CS4.2 AR9.6 OD9].osu",
            "beatmap_folder": "1271974 Eternal Tears Of Sorrow - Another Me",
            "beatmap_file": "Eternal Tears Of Sorrow - Another Me (LMT) [Insane 1.36x (250bpm) CS4.2 AR9.6 OD9].osu",
            "background_file": "night-3129908_1920.jpg",
            "background_path_full": "/path/to/osu/Songs/1271974 Eternal Tears Of Sorrow - Another Me/night-3129908_1920.jpg"
        }
    },
    "keyoverlay": {
        "k1_pressed": false,
        "k1_count": 0,
        "k2_pressed": false,
        "k2_count": 0,
        "m1_pressed": false,
        "m1_count": 0,
        "m2_pressed": false,
        "m2_count": 0
    },
    "current_bpm": 0,
    "kiai_now": false,
    "current_pp": 469.34991682892615,
    "fc_pp": 0,
    "ss_pp": 469.34991682892615,
    "menu_mods": 0,
    "mods_str": [],
    "plays": 7
}

Notes

All gameplay data is reset when leaving Playing state, so you don't need to do this manually in your counter
current_pp changes depending on state:
1. SongSelect - Shows SS pp's for currently selected map
2. Playing - gradually calculates pp's based on your progress into beatmap and gameplay mods
3. ResultScreen - calculates pp for score that appears on result screen
unstable_rate - converts itself according to gameplay mods
fc_pp - removes misses of scores when playing
ss_pp - uses mods based on your current state
mods_str changes depending on state:
1. SongSelect - uses menu mods
2. Playing - uses gameplay mods
3. ResultScreen - uses result_screen mods
current_stars - calculates stars gradualy based on your progress into beatmap and gameplay mods
playtime changes depending on state:
1. SongSelect - represents progress of mp3
2. Playing - represents your progress into current beatmap

Benchmarks (Linux)

Static addresses reading

Around ~600 ms

Reading loop

The Sun The Moon The Stars +DT (x2 replay) ~190us
Plasma Gun [Extreme] +HDDT ~90us

Development

Profiling

You can use tracy to profile rosu-memory. Just build rosu with --features tracy-profile argument and then connect to client.

Command line arguments

Options:
  -o, --osu-path       <OSU_PATH>        Path to osu! folder [env: OSU_PATH=]
  -i, --interval       <INTERVAL>        Interval between updates in ms [default: 300]
  -e, --error-interval <ERROR_INTERVAL>  Amount of seconds waiting after critical error happened before running again
  -h, --help                             Print help

486c / rosu-memory

readme