Accessories

The accessories currently include the following subprojects:

Validator API (formerly heroku)

Simple mzqc-validator API

The simple API has three endpoints:

1. to indicate ‘/status/’ (GET)

2. providing ‘/documentation/’ (GET)

3. to post mzQC files to ‘/validator/’ (POST)

The documentation endpoint provides a dict with details to each part of the validation (key) as text (value). The validator endpoint takes a mzQC file (JSON) and responds with an object as described in the documentation endpoint.

Build

From the root of the pymzqc source folder (i.e. build context pymzqc/) build the accessories/onlinevalidator/Dockerfile, e.g. with podman:

podman build -t mzqc-validator -f accessories/onlinevalidator/Dockerfile .

(If you are testing a release without pypi package uncomment the respective lines in the Dockerfile to override the pymzqc version used.)

Pre-built container images for selected (pre-)release versions can be found at the mzqc-validator container registry. If you want to deploy the onlinevalidator with your local pymzqc installation, please be aware of extra dependencies to the online-validator.

Deployment

To test a deployment, run the mzqc-validator flask app in gunicorn from the container (as described in wsgi.py).

podman run -p 5000:5000 -ti localhost/mzqc-validator python3 -m gunicorn wsgi:app -b 0.0.0.0:5000 --chdir mzqc-validator/

For local tests calling the flask app directly (i.e. as single thread app) is fine too: python accessories/onlinevalidator/mzqc_online_validator.py; note that the ports might differ, depending on the flask and system defaults. Calling the mzqc_online_validator directly in gunicorn is fine too (podman run -p 8123:8123 -ti localhost/mzqc-validator python3 -m gunicorn mzqc_online_validator:app -b 0.0.0.0:8123 --chdir mzqc-validator/), the wsgi.py indirection is a legacy effect from heroku’s Procfile use and their example app.

The validate function of SemanticCheck is considerate of the environment variable MAX_ERR which set to an integer limits the amount of validation errors that can occur before validation is aborted. This can be for example adjusted in the call like so: podman run --env 'MAX_ERR=5' -p 5000:5000 -ti localhost/mzqc-validator python3 -m gunicorn wsgi:app -b 0.0.0.0:5000 --chdir mzqc-validator/

A Docker compose deploment example can be found at accessories/onlinevalidator/compose.yaml.

Port Mapping

The mzQC gitHub-pages integration and local_validator.html expect the API to run on port 5000.

Local Customisation, Development, Testing

First use dev-test-validation.py as testbed for new changes.

Then, you can build and deploy the container as described above and access the API e.g. with accessories/onlinevalidator/local_validator.html You can find both files necessary within the repository under accessories/onlinevalidator.

Also, for either testing of a local deployment or as convenience for local validation, the repository provides a html page to call a locally deployed mzqc-validator API accessories/onlinevalidator/local_validator.html

Legacy Heroku Deployment

Or you can deploy your own heroku dyno like so:

cd /tmp/
curl https://cli-assets.heroku.com/install-ubuntu.sh | sh
heroku login
heroku git:clone -a mzqc-validator
cd mzqc-validator
rsync -aP --delete /home/walzer/psi/pymzqc/acessories/onlinevalidator  /tmp/mzqc-validator
git push heroku master