| Title: | Partial Function Application and Currying with %<%, %-<%, %><%, and %<!% |
|---|---|
| Description: | Partial application is the process of reducing the arity of a function by fixing one or more arguments, thus creating a new function lacking the fixed arguments. The curry package provides three different ways of performing partial function application by fixing arguments from either end of the argument list (currying and tail currying) or by fixing multiple named arguments (partial application). This package provides this functionality through the %<%, %-<%, and %><% operators which allows for a programming style comparable to modern functional languages. Compared to other implementations such a purrr::partial() the operators in curry composes functions with named arguments, aiding in autocomplete etc. |
| Authors: | Thomas Lin Pedersen |
| Maintainer: | Thomas Lin Pedersen <[email protected]> |
| License: | GPL (>=2) |
| Version: | 0.1.1 |
| Built: | 2024-09-20 12:54:02 UTC |
| Source: | https://github.com/thomasp85/curry |
The curry function and the %<% operator performs currying on
a function by partially applying the first argument, returning a function
that accepts all but the first arguments of the former function. If the first
argument is ... the curried argument will be interpreted as part of
the ellipsis and the ellipsis will be retained in the returned function. It
is thus possible to curry functions comtaining ellipis arguments to infinity
(though not adviced).
fun %<% arg curry(fun, arg)fun %<% arg curry(fun, arg)
fun |
A function to be curried. Can be any function (normal, already curried, primitives). |
arg |
The value that should be applied to the first argument. |
A function with the same arguments as fun except for the
first, unless the first is ... in which case it will be retained.
Multiple currying does not result in multiple nested calls, so while the first currying adds a layer around the curried function, potentially adding a very small performance hit, currying multiple times will not add to this effect.
Other partials: partial,
tail_curry
# Equivalent to curry(`+`, 5) add_5 <- `+` %<% 5 add_5(10) # ellipsis are retained when currying bind_5 <- cbind %<% 5 bind_5(1:10)# Equivalent to curry(`+`, 5) add_5 <- `+` %<% 5 add_5(10) # ellipsis are retained when currying bind_5 <- cbind %<% 5 bind_5(1:10)
The partial function and the %><% operator allows you to
partially call a function with a list of arguments. Named elements in the
list will be matched to function arguments and these arguments will be
removed from the returned function. Unnamed elements are only allowed for
functions containing an ellipsis, in which case they are considered part of
the ellipsis.
fun %><% args partial(fun, args)fun %><% args partial(fun, args)
fun |
A function to be partially applied. Can be any function (normal, already partially applied, primitives). |
args |
A list of values that should be applied to the function. |
A function with the same arguments as fun except for the ones
given in args
Multiple partial application does not result in multiple nested calls, so while the first partial call adds a layer around the called function, potentially adding a very small performance hit, partially calling multiple times will not add to this effect.
Other partials: curry,
tail_curry
dummy_lengths <- vapply %><% list(FUN = length, FUN.VALUE = integer(1)) test_list <- list(a = 1:5, b = 1:10) dummy_lengths(test_list)dummy_lengths <- vapply %><% list(FUN = length, FUN.VALUE = integer(1)) test_list <- list(a = 1:5, b = 1:10) dummy_lengths(test_list)
The set_defaults function and the %<?% operator modifies the
defaults of a function, returning a new function. As such it can be thought
of as a soft partial application in that the arguments does not become fixed
and the arity doesn't change, but the arguments can be ignored when making
the final call.
set_defaults(fun, defaults) fun %<?% defaultsset_defaults(fun, defaults) fun %<?% defaults
fun |
A function whose argument default(s) should be changed |
defaults |
A named list of values. The values will be set as default for the arguments matching their name. Non-matching elements will be ignored |
A new function with changed defaults
testfun <- function(x = 1, y = 2, z = 3) { x + y + z } testfun() testfun2 <- testfun %<?% list(y = 10) testfun2()testfun <- function(x = 1, y = 2, z = 3) { x + y + z } testfun() testfun2 <- testfun %<?% list(y = 10) testfun2()
The curry function and %<% operator does not perform currying
in the strictest sense since it is really "just" a partial application of the
first argument. Strictly speaking currying transforms a function into a
function taking a single argument and returning a new function accepting a
new argument as long as the original function has arguments left. Once all
arguments has been consumed by function calls it evaluates the original call
and returns the result. Strict currying has less applicability in R as it
cannot work with functions containing '...' in its argument list as it will
never reach the end of the argument list and thus never evaluate the call.
Strict currying is still provided here for completeness. The Curry()
function turns a function into a curried function reducing the ariety to one.
The %<!% operator both transforms the function into a curried one
and calls it once with the first argument. Once a function is curried it is
still possible to use %<%, %-<%, and %><% though
they all performs the same operation as the function does only accept a
single argument. As with the other functions in the curry package,
argument names and defaults are retained when performing strict currying.
Calling a curried function without providing a value for it will call it with
the default or set the argument to missing.
fun %<!% arg Curry(fun)fun %<!% arg Curry(fun)
fun |
A function to be turned into a curried function. |
arg |
A value to be used when calling the curried function |
A function accepting a single argument and returing either a new function accepting a single argument, or the result of the function call if all arguments have been provided.
testfun <- function(x, y, z) { x + y + z } curriedfun <- Curry(testfun) curriedfun(1)(2)(3) # Using the operator testfun %<!% 1 %<!% 2 %<!% 3 # The strict operator is only required for the first call testfun %<!% 1 %<% 2 %<% 3testfun <- function(x, y, z) { x + y + z } curriedfun <- Curry(testfun) curriedfun(1)(2)(3) # Using the operator testfun %<!% 1 %<!% 2 %<!% 3 # The strict operator is only required for the first call testfun %<!% 1 %<% 2 %<% 3
The tail_curry function and the %-<% operator performs
currying on a function by partially applying the last argument, returning a
function that accepts all but the last arguments of the former function. If
the last argument is ... the curried argument will be interpreted as
the last named argument. If the only argument to the function is ...
the curried argument will be interpreted as part of the ellipsis and the
ellipsis will be retained in the returned function. It is thus possible to
curry functions comtaining ellipis arguments to infinity (though not
adviced).
fun %-<% arg tail_curry(fun, arg)fun %-<% arg tail_curry(fun, arg)
fun |
A function to be curried from the end. Can be any function (normal, already (tail_)curried, primitives). |
arg |
The value that should be applied to the last argument. |
A function with the same arguments as fun except for the
last named argument, unless the only one is ... in which case it will
be retained.
Multiple tail_currying does not result in multiple nested calls, so while the first tail_currying adds a layer around the curried function, potentially adding a very small performance hit, tail_currying multiple times will not add to this effect.
Other partials: curry, partial
# Equivalent to tail_curry(`/`, 5) divide_by_5 <- `/` %-<% 5 divide_by_5(10) no_factors <- data.frame %-<% FALSE no_factors(x = letters[1:5])# Equivalent to tail_curry(`/`, 5) divide_by_5 <- `/` %-<% 5 divide_by_5(10) no_factors <- data.frame %-<% FALSE no_factors(x = letters[1:5])