pip3 install .or for editable:
pip3 install --prefix=$(python3 -m site --user-base) --editable .The easiest way to use this library is to use the dockrice.argparse.ArgumentParser instant in your python script:
import os
check_name = "AM_I_IN_A_DOCKER"
if os.getenv(check_name, None) is None:
import dockrice.argparse as argparse
parser_kwargs = {
"image": "python",
"docker_kwargs": {"environment": {check_name: ""}},
}
else:
import argparse
parser_kwargs = {}
parser = argparse.ArgumentParser(
description="A simple example of running a script in docker.", **parser_kwargs
)
# add more arguments here, use pathlib.Path for any file/directory
args = parser.parse_args()
# add additional imports and code here
This works because when the module is loaded, the environmental variable AM_I_IN_A_DOCKER is not defined. Thus, it uses the argparser from dockrice, with the respective parameters defined. For any file/directory that you use in your script you need an argument in the argparser with type=pathlib.Path. The argparser then starts a image, mounts all necessary path and runs the script in the container. The AM_I_IN_A_DOCKER variable is defined inside the container as it was added to the environmental variables passed to the docker run command, so it uses the normal argparse module and thus will run the full code without any special consideration.
The above code is thus about equivalent to
docker run -v $LOCAL_DATA_DIR:/data my_container python my_script.py --data-dir /data
but you only have to type
python my_script.py --data-dir $LOCAL_DATA_DIR
and dockrice takes care of the rest for you.
The libary also offers a class DockerPath. It inherits from pathlib.Path, and can be used like any pathlib.Path object. However, it additionally also keeps track of the path a file or directory will be mounted inside a docker. This can be accessed with the mount_path method or you can get a docker.types.Mount object that can be used to correctly mount the file through the get_mount method. The mount_path is a pathlib.PurePosixPath, thus always represents the path of a linux file, while the DockerPath supports both windows and linux path syntax.