Automation for the incremental idle game Your Chronicle. https://store.steampowered.com/app/1546320/Your_Chronicle/
This code isn't really meant to be run as-is. It's meant as an example or starting point for anyone else who's interested in automation for Your Chronicle or automation in general.
Completing ED1 twice in a row: https://youtu.be/MNwzsI3IOA8
- Linux. Tested on Debian 12, but others should work.
sudo apt install wmctrl xdotool scrot
This was developed on Debian 12. It should work for other linux
distros without modification. With modifications to the programs used in
calls to subprocess.run as well updates to run.sh, it should work for
non-linux systems.
The programs used through subprocess.run are:
wmctrlfor moving and resizing the game window.xdotoolfor verifying that the "Your Chronicle" window is active.scrotfor capturing screenshots. The screenshot function from pyautogui is not reliable. https://gist.github.com/Omustardo/25d14d0bcc8acf3c979b6505c2f136c7
git clone https://github.com/omustardo/auto-yourchronicle.git
cd auto-yourchronicle
chmod +x run.sh
./run.shThe first time it runs, it will download all of the python dependencies
documented in src/requirements.txt.
Start by making sure you can run.sh as documented above in the Usage section.
This will ensure your python virtual environment is set up.
Modifying the Python code needs no explanation, and you can continue to use
run.sh to run it. If you need to add new python dependencies, you'll need to:
- Enter the virtual environment :
source ./src/venv/bin/activate - Install new deps :
pip3 install foo - Update the list of dependencies :
pip3 freeze > ./src/requirements.txt
I used OBS to record both the terminal window and the game window. It's only about 1GB/h and it's probably the most effective way to find rare issues.
Saving screenshots to disk is also a great way to find issues, and is how I detected a bug in how PIL takes screenshots: https://gist.github.com/Omustardo/25d14d0bcc8acf3c979b6505c2f136c7
Your Chronicle is very consistent about where its UI elements go, so I always
start the program by moving the game to a fixed location and give it a fixed
size. I then have a bunch of pre-set regions (x,y,w,h) that I can search for
text in. This code is all in navigator.py.
In order to navigate between menus, I keep a url-like string
(navigator.menu_state). Whenever I change to a different view, I update this.
This allows different methods to know if they need to change to a different menu
or not.
Many methods return a boolean to indicate if they worked or not. Generally if something fails it isn't recoverable and the program just exits, but in a few cases it is recoverable.
I began automation using Sikuli (http://sikulix.com/). Sikuli is an IDE that lets you write Python to automate detection of images and clicking places on the screen. Its IDE is nice because it shows images inline, so if you have code that tries to detect a certain image, it's easy to read the code. I ended up looking for alternatives because Sikuli felt limiting in a few ways:
- documentation was initially hard to find
- I wanted to run a standard python program, but instead I had to run Sikuli as a java binary and then run python code from the Sikuli IDE.
- It doesn't have OCR built in, doing anything required taking screenshots and matching against those. This meant that I couldn't resize the image, and it even seemed to have trouble with Redshift (changing the color of my monitor at night).
- It felt flaky. The program would sometimes fail to match images that it had matched before and I never figured out why.
I ended up switching to pytesseract + pyautogui and was pleased with making that change. pytesseract is an OCR tool. pyautogui moves the mouse and does other GUI interactions.
Unfortunately I started using version control well after I switched away from Sikuli, so that isn't saved in this repo.