How do i run a shell script inside docker-compose - django

Question

I have a use case in which i am trying to build a django based rest api and then use continuous integration using travis CI when changes are pushed to github. I am also using docker to build and docker-compose to scale my services. The problem is i want to run pytest and flake8 when i push my changes to github. Now i have not added any tests and hence the pytest command is giving an exit status of 5.
To get around this i tried creating a script to do this :
#!/bin/bash
pytest;
err=$? ;
if (( $err != 5 )) ;
then
exit $err;
fi
flake8 ;
But i cannot get docker-compose to run this . When i run the script using the command :
docker-compose run app sh -c "run_script.sh"
It gives the below error message :
sh: run_script.sh: not found
Below is my docker-compose yml file:
version: "3"
services:
app:
build:
context: .
ports:
- "8000:8000"
volumes:
- ./app:/app
command: >
sh -c "python manage.py runserver 0.0.0.0:8000"
And below is the dockerfile:
FROM python:3.7-alpine
MAINTAINER Subhayan Bhattacharya
ENV PYTHONUNBUFFERED 1
COPY Pipfile* /tmp/
RUN cd /tmp && pip install pipenv && pipenv lock --requirements > requirements.txt
RUN pip install -r /tmp/requirements.txt
RUN mkdir /app
WORKDIR /app
COPY ./app /app
RUN adduser -D user
USER user
This should be a simple issue but i cannot figure out how to get around this.
Can someone please help me find the solution

Answer 1

Your script isn't working because Alpine base images don't have GNU bash. Your script almost limits itself to the POSIX Shell Command Language; if you do, you can change the "shebang" line to #!/bin/sh.
#!/bin/sh
# ^^^ not bash
pytest # individual lines don't need to end with ;
err=$?
# use [ ... ] (test), not ((...))
if [ "$err" -ne 5 ] && [ "$err" -ne 0 ]; then
exit "$err"
fi
flake8
In the context of a CI system, it is important to remove the volumes: line that mounts a local directory over your container's /app directory: having that line means you are not testing what's in your image at all, but instead a possibly-related code tree that's on the host system.
In practice I'd suggest running both of these tools in a non-Docker environment. It will be simpler to run them and collect their results. Especially a style checker like flake8 will have very few dependencies on system packages or other containers being started, and ideally your unit tests can also run without hard-to-set-up context like a database container as well. I'd suggest a sequence like:
Check out the source code.
Create a virtual environment and install its dependencies.
Run pytest, flake8, and similar test tools.
Then build a Docker image, without test-only tools.
Run the image with its assorted dependencies.
Run further tests based on network calls into the container.