test – Domain-specific language for writing tests

Domain-specific language for writing numeric tests.

This module provides a few classes and functions to write numeric tests.

Let us import the relevant modules first:

>>> from collections import OrderedDict
>>> from valjean.eponine.dataset import Dataset
>>> import numpy as np

Now we create a toy data set:

>>> x = np.linspace(-5., 5., num=100)
>>> y = x**2
>>> error = np.zeros_like(y)
>>> bins = OrderedDict()
>>> bins['x'] = x
>>> parabola = Dataset(y, error, bins=bins)

We perturb the data by applying some small amount of noise:

>>> eps = 1e-8
>>> noise = np.random.uniform(-eps, eps, parabola.shape)
>>> y2 = y + noise
>>> parabola2 = Dataset(y2, error, bins=bins)

Now we can test if the new dataset is equal to the original one:

>>> from valjean.gavroche.test import TestEqual
>>> test_equality = TestEqual(parabola, parabola2, name="parabola",
...                           description="equality test")
>>> test_equality_res = test_equality.evaluate()
>>> print(bool(test_equality_res))
False

However, they are approximately equal:

>>> from valjean.gavroche.test import TestApproxEqual
>>> test_approx = TestApproxEqual(parabola, parabola2, name="parabola",
...                               description="approx equal test")
>>> test_approx_res = test_approx.evaluate()
>>> print(bool(test_approx_res))
True

exception valjean.gavroche.test.CheckBinsException

An error is raised when check bins fails.

valjean.gavroche.test.same_arrays(arr1, arr2)

Return True if arr1 and arr2 are equal.

Parameters:

• arr1 – the first array.

• arr2 – the second array.

valjean.gavroche.test.same_bins(bins1, bins2)

Return True if all the coordinate arrays are compatible.

Parameters:

• bins1 – the first dictionary of coordinate arrays.

• bins2 – the second dictionary of coordinate arrays.

valjean.gavroche.test.same_bins_datasets(*datasets)

Return True if all datasets have the same coordinates.

Parameters:

datasets (Dataset) – any number of datasets.

valjean.gavroche.test.check_bins(*datasets)

Check if the datasets have compatible coordinates, raise if not.

Raises:

ValueError – if the datasets do not have compatible coordinates.

class valjean.gavroche.test.Test(*, name, description='', labels=None)

Generic class for comparing any kind of results.

Base class for tests.

__init__(*, name, description='', labels=None)

Initialize the Test object with a name, a description of the test (may be long) and labels if needed.

The test is actually performed in the evaluate method, which is abstract in the base class and must be implemented by sub-classes.

Parameters:

• name (str) – name of the test, this string will typically end up in the test report as a section name.

• description (str) – description of the test exepcted with context, this string will typically end up in the test report.

• labels (dict) – labels to be used for test classification in reports (for example category, input file name, type of result, …)

abstract evaluate()

Evaluate the test on the given inputs.

Must return a subclass of TestResult.

data()

Generator yielding objects supporting the buffer protocol that (as a whole) represent a serialized version of self.

class valjean.gavroche.test.TestDataset(dsref, *datasets, name, description='', labels=None)

Generic class for comparing datasets.

__init__(dsref, *datasets, name, description='', labels=None)

Initialisation of TestEqual:

Parameters:

• name (str) – name of the test (in analysis)

• description (str) – specific description of the test

• labels (dict) – labels to be used for test classification in reports (for example category, input file name, type of result, …)

• dsref (Dataset) – reference dataset

• datasets (list(Dataset)) – list of datasets to be compared to reference dataset

data()

Generator yielding objects supporting the buffer protocol that (as a whole) represent a serialized version of self.

abstract evaluate()

Evaluate the test on the given datasets.

Must return a subclass of TestResult.

class valjean.gavroche.test.TestResult(test)

Base class for test results.

This result should be filled by Test daughter classes.

__init__(test)

Initialisation of TestResult.

Parameters:

test (Test used) – the used test

class valjean.gavroche.test.TestResultFailed(test, msg)

Class for failed TestResults when an exception was raised during the evaluation.

__init__(test, msg)

Initialisation of TestResult.

Parameters:

test (Test used) – the used test

class valjean.gavroche.test.TestResultEqual(test, equal)

Result from TestEqual.

__init__(test, equal)

Initialisation of the result from TestEqual:

Parameters:

• test (TestEqual) – the used test

• equal (list (numpy.bool_) if datasets are numpy.generic, list (numpy.ndarray) if datasets are numpy.ndarray with dtype == bool.) – result from the test

__bool__()

Return the result of the test: True or False or raises an exception when it is not suitable.

class valjean.gavroche.test.TestEqual(dsref, *datasets, name, description='', labels=None)

Test if the datasets values are equal. Errors are ignored.

evaluate()

Evaluation of TestEqual.

Returns:

TestResultEqual

data()

Generator yielding objects supporting the buffer protocol that (as a whole) represent a serialized version of self.

class valjean.gavroche.test.TestResultApproxEqual(test, approx_equal)

Result from TestApproxEqual.

__init__(test, approx_equal)

Initialisation of the result from TestApproxEqual:

Parameters:

• test (TestApproxEqual) – the used test

• approx_equal (numpy.generic if datasets are numpy.generic, numpy.ndarray if datasets are numpy.ndarray. In both cases dtype == bool.) – result from the test

class valjean.gavroche.test.TestApproxEqual(dsref, *datasets, name, description='', labels=None, rtol=1e-05, atol=1e-08)

Test if the datasets values are equal within the given tolerances. Errors are ignored.

__init__(dsref, *datasets, name, description='', labels=None, rtol=1e-05, atol=1e-08)

Initialisation of TestApproxEqual:

Parameters:

• name (str) – local name of the test

• description (str) – specific description of the test

• labels (dict) – labels to be used for test classification in reports (for example category, input file name, type of result, …)

• dsref (Dataset) – reference dataset

• datasets (list(Dataset)) – list of datasets to be compared to reference dataset

• rtol (float) – relative tolerance, default =

• atol (float) – absolute tolerance, default =

To get more details on rtol and atol parameters, see numpy.isclose.

evaluate()

Evaluation of TestApproxEqual.

Returns:

TestResultApproxEqual

data()

Generator yielding objects supporting the buffer protocol that (as a whole) represent a serialized version of self.