Parameterized Unit Testing

Description

'patrick' (parameterized testing in R is kind of cool!) is a 'testthat' extension that lets you create reusable blocks of a test codes. Parameterized tests are often easier to read and more reliable, since they follow the DNRY (do not repeat yourself) rule. To do this, define tests with the function [with_parameters_test_that()]. Multiple approaches are provided for passing sets of cases.

Details

This package is inspired by parameterized testing packages in other languages, notably the ['parameterized'](https://github.com/wolever/parameterized) library in Python.

Author(s)

Maintainer: Michael Quinn msquinn@google.com

Other contributors:

• Michael Chirico chiricom@google.com [contributor]

See Also

Useful links:

• https://github.com/google/patrick

• Report bugs at https://github.com/google/patrick/issues

Examples

with_parameters_test_that("trigonometric functions match identities:",
  {
    testthat::expect_equal(expr, numeric_value)
  },
  expr = c(sin(pi / 4), cos(pi / 4), tan(pi / 4)),
  numeric_value = c(1 / sqrt(2), 1 / sqrt(2), 1),
  .test_name = c("sin", "cos", "tan")
)

# Run the same test with the cases() constructor
with_parameters_test_that(
  "trigonometric functions match identities",
  {
    testthat::expect_equal(expr, numeric_value)
  },
  cases(
    sin = list(expr = sin(pi / 4), numeric_value = 1 / sqrt(2)),
    cos = list(expr = cos(pi / 4), numeric_value = 1 / sqrt(2)),
    tan = list(expr = tan(pi / 4), numeric_value = 1)
  )
)

# If names aren't provided, they are automatically generated.
with_parameters_test_that(
  "trigonometric functions match identities",
  {
    testthat::expect_equal(expr, numeric_value)
  },
  cases(
    list(expr = sin(pi / 4), numeric_value = 1 / sqrt(2)),
    list(expr = cos(pi / 4), numeric_value = 1 / sqrt(2)),
    list(expr = tan(pi / 4), numeric_value = 1)
  )
)
# The first test case is named "expr=0.7071068, numeric_value="0.7071068"
# and so on.

# Or, pass a data frame of cases, perhaps using a helper function
make_cases <- function() {
  tibble::tribble(
    ~.test_name, ~expr, ~numeric_value,
    "sin", sin(pi / 4), 1 / sqrt(2),
    "cos", cos(pi / 4), 1 / sqrt(2),
    "tan", tan(pi / 4), 1
  )
}

with_parameters_test_that(
  "trigonometric functions match identities",
  {
    testthat::expect_equal(expr, numeric_value)
  },
  .cases = make_cases()
)

Execute a test with parameters.

Description

This function is an extension of [testthat::test_that()] that lets you pass a series of testing parameters. These values are substituted into your regular testing code block, making it reusable and reducing duplication.

Usage

with_parameters_test_that(
  desc_stub,
  code,
  ...,
  .cases = NULL,
  .test_name = NULL,
  .interpret_glue = TRUE
)

cases(...)

Arguments

Details

You have a couple of options for passing parameters to you test. You can use named vectors/ lists. The function will assert that you have correct lengths before proceeding to test execution. Alternatively you can used a 'data.frame' or list in combination with the splice unquote operator !!!. Last, you can use the constructor 'cases()', which is similar to building a 'data.frame' rowwise. If you manually build the data frame, pass it in the '.cases' argument.

## Naming test cases

If the user passes a character vector as '.test_name', each instance is combined with 'desc_stub' to create the completed test name. Similarly, the named argument from 'cases()' is combined with 'desc_stub' to create the parameterized test names. When names aren't provided, they will be automatically generated using the test data.

Names follow the pattern of "name=value, name=value" for all elements in a test case.

Examples

with_parameters_test_that("trigonometric functions match identities:",
  {
    testthat::expect_equal(expr, numeric_value)
  },
  expr = c(sin(pi / 4), cos(pi / 4), tan(pi / 4)),
  numeric_value = c(1 / sqrt(2), 1 / sqrt(2), 1),
  .test_name = c("sin", "cos", "tan")
)

# Run the same test with the cases() constructor
with_parameters_test_that(
  "trigonometric functions match identities",
  {
    testthat::expect_equal(expr, numeric_value)
  },
  cases(
    sin = list(expr = sin(pi / 4), numeric_value = 1 / sqrt(2)),
    cos = list(expr = cos(pi / 4), numeric_value = 1 / sqrt(2)),
    tan = list(expr = tan(pi / 4), numeric_value = 1)
  )
)

# If names aren't provided, they are automatically generated.
with_parameters_test_that(
  "trigonometric functions match identities",
  {
    testthat::expect_equal(expr, numeric_value)
  },
  cases(
    list(expr = sin(pi / 4), numeric_value = 1 / sqrt(2)),
    list(expr = cos(pi / 4), numeric_value = 1 / sqrt(2)),
    list(expr = tan(pi / 4), numeric_value = 1)
  )
)
# The first test case is named "expr=0.7071068, numeric_value="0.7071068"
# and so on.

# Or, pass a data frame of cases, perhaps using a helper function
make_cases <- function() {
  tibble::tribble(
    ~.test_name, ~expr, ~numeric_value,
    "sin", sin(pi / 4), 1 / sqrt(2),
    "cos", cos(pi / 4), 1 / sqrt(2),
    "tan", tan(pi / 4), 1
  )
}

with_parameters_test_that(
  "trigonometric functions match identities",
  {
    testthat::expect_equal(expr, numeric_value)
  },
  .cases = make_cases()
)