QGIS/tests

QGIS unit tests

Build tests

Make sure that you have enabled building of tests in CMake. cmake -DENABLE_TESTS=ON ..

Run tests

You can run all tests using make check. Note you will need xvfb-run for that (sudo apt-get install xfvb).

Individual tests can be run using ctest.

For example if the output of make check ends like this:

   The following tests FAILED:
         77 - PyQgsLocalServer (Failed)

You could re-run the failing test with:

   ctest -V -R PyQgsLocalServer

The parameter -V enables verbose mode and -R takes a regular expression as parameter and will only run matching tests.

For python tests, you can run a specific test inside a unit file with something like this:

 QGIS_PREFIX_PATH=output PYTHONPATH=output/python:$PYTHONPATH \
   python ${srcdir}/tests/src/python/test_qgsvectorfilewriter.py
   TestQgsVectorLayer.testOverwriteLayer

If you get Could not connect to any X display errors it means that your build machine does not have an X server. In that case you need to run the test under xvfb-run. For example:

    xvfb-run --server-args=-screen\ 10\ 1024x768x24 ctest -V -R PyQgsServerWMSGetMap

Advanced configuration

Postgres

Make sure that you have enabled building of postgres test in CMake. cmake -DENABLE_TESTS=ON -DENABLE_PGTEST=ON ..

To test the postgres provider you will need to have a database available to which the postgres provider can connect. The server will need to have PostGIS support enabled.

By default the test uses one of the following connection string:

dbname=qgis_test
service=qgis_test

If these do not match your setup you can set the environment variable QGIS_PGTEST_DB to the desired connection string. Note that you can rely on standard libpq environment variables to tweak host, port user and password (PGHOST, PGPORT, PGUSER, PGPASSWORD).

Please note that the test database needs to be initialized using the sql-scripts:

tests/testdata/provider/testdata_pg*.sql

They take care of activating PostGIS for the test database and create some tables containing test data.

For convenience, a shell script is provided to create the database and initialize it as needed:

tests/testdata/provider/testdata_pg.sh

Write tests

Instructions about writing tests for the processing framework can be found in a separate README file:

${TOP_SRCDIR}/python/plugins/processing/tests/README.md

Information about labeling tests design and organization:

${TOP_SRCDIR}/tests/testdata/labeling/README.rst

WCS testing information can be found in:

${TOP_SRCDIR}/tests/testdata/raster/README.WCS

About benchmark tests you can read:

${TOP_SRCDIR}/tests/bench/README

Run python tests in GDB

First find out the required environment variables by running the test outside the debugger.

ctest -V -R ProcessingQgisAlgorithmsTest

Which prints for somewhere in the initialization code something like:

export LD_LIBRARY_PATH=NOTFOUND:/home/m-kuhn/dev/cpp/qgis/build-qt5/output/lib:
export PYTHONPATH=/home/m-kuhn/dev/cpp/qgis/build-qt5/output/python/:/home/m-kuhn/dev/cpp/qgis/build-qt5/output/python/plugins:/home/m-kuhn/dev/cpp/qgis/QGIS/tests/src/python:

First, run these two commands in the terminal.

On the following line it says something like:

-- Running /usr/bin/python3 /home/m-kuhn/dev/cpp/qgis/QGIS/python/plugins/processing/tests/QgisAlgorithmsTest.py

Which you can run in gdb with:

gdb -ex r --args /usr/bin/python3 /home/m-kuhn/dev/cpp/qgis/QGIS/python/plugins/processing/tests/QgisAlgorithmsTest.py

Now you can start using the usual gdb (bt etc.) interface or - if you have installed the appropriate debug tools (adjust for python3!) even allows doing python introspection (py-bt).