This docker image is a base for Satel webapp's python backend. The standard packages we use in most projects are already installed. The folder structure and commands provide the standard framework for development and deployment to production of the app.
There is a dev.sh
script included that builds both docker images and runs
one of 3 cases based on the parameter you pass in:
./dev.sh validate
runs the validation scripts (validatecodeonce)./dev.sh run
runs the main application../dev.sh cli
drops you into the cli for the image
The version 4 comes with significant changes:
- The docker base image was changed from alpine to slim buster
- All the
apk add
commands need to be replaced with equivalentapt-get
commands - slim buster has more base packages such as bash so probably less installation of system packages is needed
- All the
- The entrypoint now provides standard commands to run and test the webapp
- No
CMD
is necessary for production asstartapp
is the default - In development, run the
developapp
command - See below for testing
- No
- A
config.sh
is expected where environment variables for configuration purposes can be defined - The
requirements.txt
has been reduced to the base and development python packages- Base packages such as FastAPI, pydantic, loguru, ...
- Development pacakages such as watchmedo, pytest, mypy, ...
- Therefore you need to add to your requirements file packages such as pandas that were included in this image in versions prior to 4
The /python
folder is the base folder for all the files of this images
such as the entrypoint.sh
. Then the subfolders organize the app files:
/python/app
holds the source of the app/python/static
holds the static files served by the app such as static html files/python/logs
is meant for the log files generated by the app/python/files
is for files uploaded through the app
- Go to docker hub to get the latest patch version for the python docker image
- Then 3 files need to be edited with the latest python minor or patch versions:
docker-compose.generate_requirements.yml
to generate the requirements files..github/workflows/docker_image.yml
to build and push the right images to docker hub.Dockerfile
to change the default base image version. This is used only for testing while developing but better to keep it up to date.
Rather simple. Just run the following command at the root of the repo:
docker-compose -f docker-compose.generate_requirements.yml up
The entrypoint of this docker image provides standard commands for our app's docker
image to call with CMD
in the Dockerfile.
These commands are effectively run as arguments of the entrypoint such as:
/python/entrypoint.sh startapp
The .bashrc
file contains aliases so that one can also just call startapp
from within the docker container.
NOTE: If you want to run a bash command, just pass it to the docker container
and it will be executed since it doesn't match one of the predefined commands.
Example here running the ls
command:
docker run satel/python-base:latest ls
Run the app in production mode. This is the default CMD
is none is provided.
The app is expected to be a FastAPI app
defined in /python/app/webapp/main.py
such that the app is executed with the command:
uvicorn --host 0.0.0.0 --port 8000 webapp.main:app
You can replace webapp.main:app
by setting the first argument of the startapp
command such that for example:
startapp myapp:theapp
The command will try to load the config.sh
from multiple locations:
- From the docker secrets folder
- From the current directory, being
/python/app
- If the previous locations don't have a
config.sh
file, theconfig.sh.example
file is loaded from the current directory if it exists as a fallback. This is useful for CI/CD environments for which the example values are good enough to run the tests
Run the app in development mode with watchmedo
from the watchdog
python package
such that the app restarts whenever the code changes in /python/app
.
The same default location (webapp.main:app
) and option to change the location
as startapp
are used for developapp
.
The configuration file is loaded the same way as in startapp
.
This command is used during development to automatically run the pytest
tests,
the mypy
typing check and the flake8
linting check whenever the code changes.
The flake8
linting includes isort
import module sorting thanks to the flake8-isort
plugin.
This code validation command executes the /python/test_suite.sh
script which can
be overwritten with custom code validation.
Same as validatecode
but executed once rather than continously running on
code change.
This can be used for CI/CD purposes since it generates report files in the /python/reports
folder.
DEPRECATED: Use validatecodeonce
instead
This command runs only the pytest
tests for CI/CD purposes. It outputs the tests
and code coverage results in the /python/app/unittesting.xml
and
/python/app/coverage.xml
files respectively.
A coverage configuration file
can be provided at python/app/coverage.conf
.