A very simple python program, suppose current directory is /PYTHON, I want to pass file.txt as argument to python script boot.py, here is my Dockerfile
FROM python
COPY boot.py ./
COPY file.txt ./
RUN pip install numpy
CMD ["python", "boot.py", "file.txt"]
then I build the Docker container with :
docker build -t boot/latest .
then run the container
docker run -t boot:latest python boot.py file.txt
I got the correct results.
But If I copy another file file1.txt to the current directory (from a different directory (not /PYTHON)), then I run the container again:
docker run -t boot:latest python boot.py file1.txt
I got the following error:
FileNotFoundError: [Errno 2] No such file or directory: 'file1.txt'
so the error is due to fact that file1.txt is not in the container, but if I share this container with a friend and the friend wants to pass a very different file as argument, how do I write the Dockerfile so anybody with my container can pass very different files as argument without errors ? Thanks in advance. (I am new to Docker)
You may also make your script read from
STDIN
and then pass data to docker usingcat
. Have a look at how to get docker container to read from stdin?Something like:
It won't work that way. Like you said, file1.txt is not in the container.
The work around is to use Docker volumes to inject files from your host machine to the container when running it.
Something like this :
/path/to/file1 would be the path on your host machine which will override /path/to/file.txt on the container.
So your friends can play on /path/to/file1 on their host machines, with their own files.
If I understand the question correctly, you are acknowledging that the file isn't in the container, and you are asking how to best share you container with the world, allowing people to add their own content into it.
You have a couple of options, either use docker volumes, which allows your friends (and other interested parties) to mount local volumes inside your docker containers. That is, you can overlay a folder on your local filesystem onto a folder inside the container(This is generally quite nifty when you are developing locally as well)
Or, again, depending on the purpose of your container, somebody could extend your image. For example, a Dockerfile like
Choose which ever option suits your project the best.