Tests

To run tests, the project needs to set up:

  • a folder tree with various geospatial formats

  • a PostGIS database

  • a uData instance to run publish operations

Commands below are meant to be used on Linux Bash but can be adapted to your own operating system.

System requirements

  • Git

  • Docker >= 22

  • DicoGIS locally cloned

Testing setup

In your virtual environment:

  1. Install development requirements (Ubuntu or Windows)

  2. Install data fixtures:

    git clone --depth=1 https://github.com/qgis/QGIS-Training-Data.git ./tests/fixtures/qgisdata
python -m pip install -U gisdata -t ./tests/fixtures

  3. Launch full composition (PostGIS + uData):

    docker compose -f "tests/container/docker-compose.dev.yml" up -d --build

It’s also possible to launch only PostGIS container:

docker compose -f "tests/container/docker-compose.dev.yml" up -d --build 'postgis'

Try it out

Inventory

On files:

dicogis-cli inventory --verbose --input-folder tests/fixtures

On PostGIS:

PGSERVICEFILE="./tests/fixtures/database/pg_service.conf"  dicogis-cli inventory --verbose --pg-services dicogis_test

Tip

You can also connect to the database through QGIS during the container life cycle:

  1. Recommended: create a QGIS profile dedicated to DicoGIS:

    qgis --profile dicogis

  2. In QGIS settings, add an environment variable PGSERVICEFILE pointing to the file under tests/fixtures/database/pg_service.conf:

    QGIS - PGSERVICEFILE variable

  3. Restart QGIS

  4. Create a connection with the following fields:

    QGIS - PosgreSQL connection for DicoGIS testing

  5. You can interact with the database:

    QGIS - PostgreSQL connection for DicoGIS listing

Publish

  1. Run an inventory with udata output:

    PGSERVICEFILE="./tests/fixtures/database/pg_service.conf" dicogis-cli inventory --language FR --output-format udata --pg-services dicogis_test --output-path ./tests/fixtures/tmp/udata

  2. Publish it to the local instance:

dicogis-cli publish --input-folder ./tests/fixtures/tmp/udata  --udata-organization-id  OOOOOOOOOOOOOOOO --udata-api-url-base http://localhost:7010/api/ --udata-api-key XXXXXXXXXXXXXXXXX

Run tests suite

pytest

Publish CI tests

Two separate CI checks cover uData publication:

  • tests/test_cli_publish_udata.py (run as part of the regular pytest suite): exercises dicogis-cli publish’s logic against a mocked uData HTTP API (using the responses library), so it runs fast without any real infrastructure.

  • .github/workflows/tester_udata_publish.yml: a real end-to-end integration test, matrixed across uData’s current and previous major versions. It builds a minimal, version-pinned uData image from udata/system (see tests/container/udata-pinned/, since the udata/udata:latest Docker Hub image doesn’t reliably expose which udata release it bundles), boots it with MongoDB and Redis (tests/container/docker-compose.ci-udata.yml), bootstraps an admin user and API token, then runs a real dicogis-cli inventory --output-format udata + dicogis-cli publish and checks the dataset actually appears in the instance.