tidyselect (version 1.1.0)

eval_rename: Evaluate an expression with tidyselect semantics

Description

eval_select() and eval_rename() evaluate defused R code (i.e. quoted expressions) according to the special rules of the tidyselect syntax. They power functions like dplyr::select(), dplyr::rename(), or tidyr::pivot_longer().

See the Get started vignette to learn how to use eval_select() and eval_rename() in your packages.

Usage

eval_rename(
  expr,
  data,
  env = caller_env(),
  ...,
  strict = TRUE,
  name_spec = NULL
)

eval_select( expr, data, env = caller_env(), ..., include = NULL, exclude = NULL, strict = TRUE, name_spec = NULL, allow_rename = TRUE )

Arguments

expr

Defused R code describing a selection according to the tidyselect syntax.

data

A named list, data frame, or atomic vector. Technically, data can be any vector with names() and "[[" implementations.

env

The environment in which to evaluate expr. Discarded if expr is a quosure.

...

These dots are for future extensions and must be empty.

strict

If TRUE, out-of-bounds errors are thrown if expr attempts to select or rename a variable that doesn't exist. If FALSE, failed selections or renamings are ignored.

name_spec

A name specification describing how to combine or propagate names. This is used only in case nested c() expressions like c(foo = c(bar = starts_with("foo"))). See the name_spec argument of vctrs::vec_c() for a description of valid name specs.

include, exclude

Character vector of column names to always include or exclude from the selection.

allow_rename

If TRUE (the default), the renaming syntax c(foo = bar) is allowed. If FALSE, it causes an error. This is useful to implement purely selective behaviour.

Value

A named vector of numeric locations, one for each of the selected elements.

The names are normally the same as in the input data, except when the user supplied named selections with c(). In the latter case, the names reflect the new names chosen by the user.

A given element may be selected multiple times under different names, in which case the vector might contain duplicate locations.

Details

The select and rename variants take the same types of inputs and have the same type of return value. However eval_rename() has a few extra constraints. It requires named inputs, and will fail if a data frame column is renamed to another existing column name. See the selecting versus renaming section in the syntax vignette for a description of the differences.

See Also

https://tidyselect.r-lib.org/articles/syntax.html or vignette("syntax", package = "tidyselect") for a technical description of the rules of evaluation.

Examples

Run this code
# NOT RUN {
library(rlang)

# Interpret defused code as selection:
x <- expr(mpg:cyl)
eval_select(x, mtcars)

# Interpret defused code as a renaming selection. All inputs must
# be named within `c()`:
try(eval_rename(expr(mpg), mtcars))
eval_rename(expr(c(foo = mpg)), mtcars)


# Within a function, use `enquo()` to defuse one argument:
my_function <- function(x, expr) {
  eval_select(enquo(expr), x)
}

# If your function takes dots, evaluate a defused call to `c(...)`
# with `expr(c(...))`:
my_function <- function(.x, ...) {
  eval_select(expr(c(...)), .x)
}

# If your function takes dots and a named argument, use `{{ }}`
# inside the defused expression to tunnel it inside the tidyselect DSL:
my_function <- function(.x, .expr, ...) {
  eval_select(expr(c({{ .expr }}, ...)), .x)
}

# Note that the trick above works because `expr({{ arg }})` is the
# same as `enquo(arg)`.


# The evaluators return a named vector of locations. Here are
# examples of using these location vectors to implement `select()`
# and `rename()`:
select <- function(.x, ...) {
  pos <- eval_select(expr(c(...)), .x)
  set_names(.x[pos], names(pos))
}
rename <- function(.x, ...) {
  pos <- eval_rename(expr(c(...)), .x)
  names(.x)[pos] <- names(pos)
  .x
}

select(mtcars, mpg:cyl)
rename(mtcars, foo = mpg)
# }

Run the code above in your browser using DataCamp Workspace