Tests live in this directory and use pytest. A shared test/conftest.py provides session-wide fixtures and config defaults. Run them with:

pytest -n auto          # in an activated virtual environment
uv run pytest -n auto   # otherwise

Most of the older suite subclasses PicardTestCase (test/picardtestcase.py), which provides helpers such as set_config_values(), patch_tagger_instance() and mktmpdir(). Newer subsystems (test/session/, test/script_text_edit/, the custom columns and theme tests) are written as plain pytest functions with fixtures. Both are supported. Match the style of the file you are editing, and do not convert existing tests from one style to the other as part of an unrelated change.

@pytest.mark.parametrize cannot be used on PicardTestCase subclasses — pytest does not inject the parameter, so the test errors with a missing-argument TypeError. Use the subtest_cases decorator instead, which reads like parametrize but injects the arguments itself, running each case in its own subTest so that one failing case neither hides the others nor stops the run:

from test.picardtestcase import (
    PicardTestCase,
    subtest_cases,
)


class ScriptTest(PicardTestCase):
    @subtest_cases(
        "expression,expected",
        [
            ("$gt(10,4)", "1"),
            ("$gt(6,6)", ""),
        ],
    )
    def test_gt(self, expression, expected):
        self.assertScriptResultEquals(expression, expected)

argvalues may also be a mapping of label -> args, for cases whose arguments do not describe themselves:

class SatisfiedTest(PicardTestCase):
    @subtest_cases(
        "overrides,expected",
        {
            'tags want all images': ({'embed_only_front': False}, False),
            'both want only the front': ({}, True),
        },
    )
    def test_satisfied(self, overrides, expected):
        self.assertEqual(is_satisfied(overrides), expected)

Unlike parametrize this stays a single test as far as pytest is concerned, so an individual case cannot be selected with -k.

Plain-pytest test files use @pytest.mark.parametrize directly; subtest_cases is only needed for PicardTestCase subclasses.

A session-scoped qapp fixture in conftest.py provides a QApplication instance that lives for the entire test run. Tests that create Qt widgets must request it as a dependency:

@pytest.fixture()
def my_widget(qapp):
    from picard.ui.widgets.mywidget import MyWidget

    return MyWidget()

Do not create your own QApplication or QCoreApplication in tests — doing so causes crashes when pytest-randomly reorders tests. Always reuse the shared qapp fixture. If a unittest-based test needs a Qt event loop, call QCoreApplication.instance() and only create a new one if it returns None.