checkmate: qassert – R documentation

Become an expert in R — Interactive courses, Cheat Sheets, certificates and more!

qassert

Quick argument checks on (builtin) R types

Description

The provided functions parse rules which allow to express some of the most frequent argument checks by typing just a few letters.

Usage

qassert(x, rules, .var.name = vname(x))

qtest(x, rules)

qexpect(x, rules, info = NULL, label = vname(x))

Arguments

`x`	[any] Object the check.
`rules`	[`character`] Set of rules. See details.
`.var.name`	[`character(1)`] Name of the checked object to print in error messages. Defaults to the heuristic implemented in `vname`.
`info`	[character(1)] Extra information to be included in the message for the testthat reporter. See `expect_that`.
`label`	[`character(1)`] Name of the checked object to print in messages. Defaults to the heuristic implemented in `vname`.

Details

The rule is specified in up to three parts.

Class and missingness check. The first letter is an abbreviation for the class. If it is provided uppercase, missing values are prohibited. Supported abbreviations:

`[bB]`	Bool / logical.
`[iI]`	Integer.
`[xX]`	Integerish (numeric convertible to integer, see `checkIntegerish`).
`[rR]`	Real / double.
`[cC]`	Complex.
`[nN]`	Numeric (integer or double).
`[sS]`	String / character.
`[fF]`	Factor
`[aA]`	Atomic.
`[vV]`	Atomic vector (see `checkAtomicVector`).
`[lL]`	List. Missingness is defined as `NULL` element.
`[mM]`	Matrix.
`[dD]`	Data.frame. Missingness is checked recursively on columns.
`[pP]`	POSIXct date.
`[e]`	Environment.
`[0]`	`NULL`.
`[*]`	placeholder to allow any type.

Note that the check for missingness does not distinguish between NaN and NA. Infinite values are not treated as missing, but can be caught using boundary checks (part 3).

Length definition. This can be one of

[*] any length,

[?] length of zero or one,

[+] length of at least one, or

[0-9]+ exact length specified as integer.

Preceding the exact length with one of the comparison operators =/==, <, <=, >= or > is also supported.
Range check as two numbers separated by a comma, enclosed by square brackets (endpoint included) or parentheses (endpoint excluded). For example, “[0, 3)” results in all(x >= 0 & x < 3). The lower and upper bound may be omitted which is the equivalent of a negative or positive infinite bound, respectively. By definition [0,] contains Inf, while [0,) does not. The same holds for the left (lower) boundary and -Inf. E.g., the rule “N1()” checks for a single finite numeric which is not NA, while “N1[)” allows -Inf.

Value

qassert throws an R exception if object x does not comply to at least one of the rules and returns the tested object invisibly otherwise. qtest behaves the same way but returns FALSE if none of the rules comply. qexpect is intended to be inside the unit test framework testthat and returns an expectation.

Note

The functions are inspired by the blog post of Bogumił Kamiński: http://rsnippets.blogspot.de/2013/06/testing-function-agruments-in-gnu-r.html. The implementation is mostly written in C to minimize the overhead.

Examples

# logical of length 1
qtest(NA, "b1")

# logical of length 1, NA not allowed
qtest(NA, "B1")

# logical of length 0 or 1, NA not allowed
qtest(TRUE, "B?")

# numeric with length > 0
qtest(runif(10), "n+")

# integer with length > 0, NAs not allowed, all integers >= 0 and < Inf
qtest(1:3, "I+[0,)")

# either an emtpy list or a character vector with <=5 elements
qtest(1, c("l0", "s<=5"))

# data frame with at least one column and no missing value in any column
qtest(iris, "D+")

checkmate

Fast and Versatile Argument Checks

v2.0.0

BSD_3_clause + file LICENSE

Authors

Michel Lang [cre, aut] (<https://orcid.org/0000-0001-9754-0393>), Bernd Bischl [ctb]

Initial release