pytest-exasol-backend Plugin

The pytest plugin pytest-exasol-backend provides pytest fixtures for using various Exasol database instances in integration tests.

In particular, the plugin enables accessing both Exasol On-Prem and SaaS backends in a unified way.

This enables your integration tests to use either backend variant without modification, incl. iterating the test for each backend variant.

Features

• Provides session level fixtures that can be turned into connection factories for the database and the BucketFS.
• Automatically makes the tests running on the selected backends.
• Allows selecting either or both backends from the CLI that executes the pytest.
• Starts the selected backends preemptively and in parallel.

Installation

The pytest-exasol-backend plugin can be installed using pip:

pip install pytest-exasol-backend

Alternatively via poetry, recommended as dev dependency:

poetry add pytest-exasol-backend --group dev

Usage in Tests

PyExasol Connection

This test accesses the database via PyExasol and requires an additional (dev) dependency to pyexasol to be added to your project.

Note: If the pytest option --backend all is specified, then this test will run twice - once for each backend.

import pyexasol

def test_simple_sql(backend_aware_database_params):
    with pyexasol.connect(**backend_aware_database_params) as conn:
        value = conn.execute('SELECT 1 FROM DUAL').fetchval()
        assert value == 1

The following test accesses a specific table within a database schema, and requires

• Database schema MY_SCHEMA and table MY_TABLE to exist
• Database table MY_TABLE to contain 5 rows

import pyexasol

def test_number_of_rows_in_my_table(backend_aware_database_params):
    with pyexasol.connect(**backend_aware_database_params, schema='MY_SCHEMA') as conn:
        num_of_rows = conn.execute('SELECT COUNT(*) FROM MY_TABLE;').fetchval()
        assert num_of_rows == 5

BucketFS

This test accesses the BucketFS. Again, this test will run for each backend, unless one of them is disabled in the CLI.

import exasol.bucketfs as bfs

def test_my_file_exists(backend_aware_bucketfs_params):
    my_bfs_dir = bfs.path.build_path(**backend_aware_bucketfs_params, path='MY_BFS_PATH')
    my_bfs_file = my_bfs_dir / 'my_file.dat'
    assert my_bfs_file.exists()

Inspect the Selected Backend Variant

For inquiring the currently selected backend variant in a test case, you can use the backend fixture, as shown below.

def test_something_backend_sensitive(backend):
    if backend == 'onprem':
        # Do something special for the On-Prem database.
        pass
    elif backend == 'saas':
        # Do something special for the SaaS database.
        pass
    else:
        raise RuntimeError(f'Unknown backend {backend}')

Selecting Backends on the Command Line

There is no default backend specified for testing.

Please use the --backend option to specify the target backend with either onprem, saas, or all.

The plugin automatically starts the selected backends and shuts them down after the test session has finished.

• A SaaS backend is started via saas-api-python.
• An On-Prem backend via the ITDE.
• Additionally, you can use an external or local database.

Please note that all selected backends are started preemptively, regardless of their actual usage in tests.

Therefore, it is important to make sure the backends are not selected if they are not needed, for instance when running unit tests only.

Example Command Lines

Run the tests on an On-Prem database:

pytest --backend=onprem my_test_suite.py

Run the tests on two backends:

pytest --backend=onprem --backend=saas my_test_suite.py

Run the test on all backends—equivalent to the previous command:

pytest --backend=all my_test_suite.py

(Re-)Using an External or Local Database

During development you can shorten the time between code changes and running the tests by (re-)using a backend that is already running.

To save 2-20 minutes each cycle, simply add CLI parameter --itde-db-version=external.

Alternatively, you can export environment variable PYTEST_ADDOPTS, e.g.

export PYTEST_ADDOPTS="--backend=onprem --itde-db-version=external"

More parameters may be required if your setup deviates from the default values:

Setting ITDE Parameters via CLI Options

Sometimes, the default ITDE parameters cannot satisfy the test requirements. The plugin allows setting some of the parameters of the api.spawn_test_environment(...) function. The parameter values can be provided in the CLI options. Currently, it is possible to set values of the following parameters:

• --itde-db-mem-size
• --itde-db-disk-size
• --itde-nameserver
• --itde-additional-db-parameter
• --itde-db-version

This example runs the tests using an instance of the DockerDB with increased memory.

pytest --backend=onprem --itde-db-mem-size "8 GiB" my_test_suite.py

These options are ignored if the "onprem" backend is not selected.

Naming SaaS instances

For naming SaaS instances, PYTBE defines fixtures database_name and project_short_tag.

Fixture database_name

• Depends on fixture project_short_tag() and
• Passes the result to SAPIPY's function timestamp_name(), see SAPIPY user guide for more details.

You can override this fixture, but this is not recommended.

Fixture project_short_tag tries to read the current project's short tag from the following places

• CLI option --project-short-tag
• Environment variable PROJECT_SHORT_TAG
• Yaml file error_code_config.yml