Skip to contents

Require usage of startsWith() and endsWith() over grepl()/substr() versions

Source: R/string_boundary_linter.R
string_boundary_linter.Rd

startsWith() is used to detect fixed initial substrings; it is more readable and more efficient than equivalents using grepl() or substr(). c.f. startsWith(x, "abc"), grepl("^abc", x), substr(x, 1L, 3L) == "abc".

Usage

string_boundary_linter(allow_grepl = FALSE)

Arguments

allow_grepl

Logical, default FALSE. If TRUE, usages with grepl() are ignored. Some authors may prefer the conciseness offered by grepl() whereby NA input maps to FALSE output, which doesn't have a direct equivalent with startsWith() or endsWith().

Details

Ditto for using endsWith() to detect fixed terminal substrings.

Note that there is a difference in behavior between how grepl() and startsWith() (and endsWith()) handle missing values. In particular, for grepl(), NA inputs are considered FALSE, while for startsWith(), NA inputs have NA outputs. That means the strict equivalent of grepl("^abc", x) is !is.na(x) & startsWith(x, "abc").

We lint grepl() usages by default because the !is.na() version is more explicit with respect to NA handling -- though documented, the way grepl() handles missing inputs may be surprising to some users.

See also

linters for a complete list of linters available in lintr.

Tags

configurable, efficiency, readability

Examples

# will produce lints
lint(
  text = 'grepl("^a", x)',
  linters = string_boundary_linter()
)
#> ::warning file=<text>,line=1,col=7::file=<text>,line=1,col=7,[string_boundary_linter] Use !is.na(x) & startsWith(x, string) to detect a fixed initial substring, or, if missingness is not a concern, just startsWith. Doing so is more readable and more efficient.

lint(
  text = 'grepl("z$", x)',
  linters = string_boundary_linter()
)
#> ::warning file=<text>,line=1,col=7::file=<text>,line=1,col=7,[string_boundary_linter] Use !is.na(x) & endsWith(x, string) to detect a fixed terminal substring, or, if missingness is not a concern, just endsWith. Doing so is more readable and more efficient.

# okay
lint(
  text = 'startsWith(x, "a")',
  linters = string_boundary_linter()
)

lint(
  text = 'endsWith(x, "z")',
  linters = string_boundary_linter()
)

# If missing values are present, the suggested alternative wouldn't be strictly
# equivalent, so this linter can also be turned off in such cases.
lint(
  text = 'grepl("z$", x)',
  linters = string_boundary_linter(allow_grepl = TRUE)
)