This directory contains the code to test the Python code. It uses the unittest module.

The tests need a spark distribution installed locally to run. An easy way to do it is to go to the Apache Spark download page and select version 2.1.1 (May 02 2017), Pre-built for Apache Hadoop 2.7 and later.

Extract the tarball in a local directory and set the following environment variable:

export SPARK_HOME=<local-spark-directory>

One time preparation for running the python tests can be setup by runinning the following from the root Flint directory:

scripts/prepare_python_tests.sh

To run the tests issue the following command from the root Flint directory:

scripts/run_python_tests.sh

If the test setup done in the default class, SparkTestCase, does not fit the needs of a particular environment, a new class can be written. The name of the new class, say MyTestCase is then exported in the BASE_CLASS variable before the tests are run:

export FLINT_BASE_TESTCASE=<Name of new class>