Related
I am a newbee in programming, and I run into an issue with R about generic function: how to write it when there are multiple inputs?
For an easy example, for dataset and function
z <- c(2,3,4,5,8)
calc.simp <- function(a,x){a*x+8}
# Test the function:
calc.simp(x=z,a=3)
[1] 14 17 20 23 32
Now I change the class of z:
class(z) <- 'simp'
How should I write the generic function 'calc' as there are two inputs?
My attempts and errors are below:
calc <- function(x) UseMethod('calc',x)
calc(x=z)
Error in calc.simp(x = z) : argument "a" is missing, with no default
And
calc <- function(x,y) UseMethod('calc',x,y)
Error in UseMethod("calc", x, y) : unused argument (y)
My confusion might be a fundamental one as I am just a beginner. Please help! Thank you very much!
I'd suggest you model your generic function off of the template used by innumerable base R functions as, e.g., mean:
> mean
function (x, ...)
UseMethod("mean")
In your case, that would translate to the following generic which (if I understand your question correctly) works just fine:
calc <- function(x, ...) UseMethod('calc')
calc.simp <- function(a, x) {
x <- unclass(x)
a * x + 8
}
## Try it out
z <- c(2,3,4,5,8)
class(z) <- "simp"
calc.simp(x = z, 10)
## [1] 28 38 48 58 88
calc(x = z, 10)
## [1] 28 38 48 58 88
When creating a function in R, we usually specify the number of argument like
function(x,y){
}
That means it takes only two arguments. But when the numbers of arguments are not specified (For one case I have to use two arguments but another case I have to use three or more arguments) how can we handle this issue? I am pretty new to programming so example will be greatly appreciated.
d <- function(...){
x <- list(...) # THIS WILL BE A LIST STORING EVERYTHING:
sum(...) # Example of inbuilt function
}
d(1,2,3,4,5)
[1] 15
You can use ... to specify an additional number of arguments. For example:
myfun <- function(x, ...) {
for(i in list(...)) {
print(x * i)
}
}
> myfun(4, 3, 1)
[1] 12
[1] 4
> myfun(4, 9, 1, 0, 12)
[1] 36
[1] 4
[1] 0
[1] 48
> myfun(4)
I have a generic function called foo. It operates differently on the classes bar and baz but there is some shared pre-processing that needs to only be done once. What is the idiomatic R way of carrying this out?
In my silly example below, I want to multiply the object being passed to the generic by 10. Yet when the method is called, the original value is printed to the console. I have read the language definition and know that the below pattern does not work. My question is: Where or how should I handle shared processing of generic arguments before calling the method?
a <- structure(1:5, class="bar")
b <- structure(6:10, class="baz")
foo <- function(x) {
x <- x * 10 # where should shared preprocessing go?
UseMethod("foo")
}
foo.bar <- function(x) {
cat("Foo!", x)
}
foo.baz <- function(x) {
cat("Baz!", x)
}
# does not propagate the operations carried out `foo`
foo(a)
foo(b)
1) Layer foo on top of actual generic Change foo to perform the preliminary code and invoke foo_ a new generic as shown. Rename foo.bar and foo.baz to foo_.bar and foo_.baz respectively so that we are left with (have also added newlines to the example):
foo <- function(x) {
x <- x * 10
foo_(x)
}
foo_ <- function(x) UseMethod("foo_")
foo_.bar <- function(x) cat("Foo!", x, "\n")
foo_.baz <- function(x) cat("Baz!", x, "\n")
Now test it out:
a <- structure(1:5, class="bar")
b <- structure(6:10, class="baz")
foo(a)
## Foo! 10 20 30 40 50
foo(b)
## Baz! 60 70 80 90 100
For an example of this in a widely used package see the source of dplyr::mutate
2) NextMethod Another way would be to give every object a class vector of two classes with "foo" made a subclass of "bar" in the case of a and of "baz" in the case of b. Then use NextMethod. Solution (1) seems simpler and it may seem weird that "foo" is a subclass of both "bar" and "baz" but here is an example of this one just in case:
foo <- function(x) UseMethod("foo")
foo.foo <- function(x) {
x <- x * 10
NextMethod()
}
foo.bar <- function(x) cat("Foo!", x, "\n")
foo.baz <- function(x) cat("Baz!", x, "\n")
Test it noting that we have changed the definitions of a and b so that they work with this approach:
a <- structure(1:5, class= c("foo", "bar"))
b <- structure(6:10, class = c("foo", "baz"))
foo(a)
## Foo! 10 20 30 40 50
foo(b)
## Baz! 60 70 80 90 100
I want to assign multiple variables in a single line in R. Is it possible to do something like this?
values # initialize some vector of values
(a, b) = values[c(2,4)] # assign a and b to values at 2 and 4 indices of 'values'
Typically I want to assign about 5-6 variables in a single line, instead of having multiple lines. Is there an alternative?
I put together an R package zeallot to tackle this very problem. zeallot includes an operator (%<-%) for unpacking, multiple, and destructuring assignment. The LHS of the assignment expression is built using calls to c(). The RHS of the assignment expression may be any expression which returns or is a vector, list, nested list, data frame, character string, date object, or custom objects (assuming there is a destructure implementation).
Here is the initial question reworked using zeallot (latest version, 0.0.5).
library(zeallot)
values <- c(1, 2, 3, 4) # initialize a vector of values
c(a, b) %<-% values[c(2, 4)] # assign `a` and `b`
a
#[1] 2
b
#[1] 4
For more examples and information one can check out the package vignette.
There is a great answer on the Struggling Through Problems Blog
This is taken from there, with very minor modifications.
USING THE FOLLOWING THREE FUNCTIONS
(Plus one for allowing for lists of different sizes)
# Generic form
'%=%' = function(l, r, ...) UseMethod('%=%')
# Binary Operator
'%=%.lbunch' = function(l, r, ...) {
Envir = as.environment(-1)
if (length(r) > length(l))
warning("RHS has more args than LHS. Only first", length(l), "used.")
if (length(l) > length(r)) {
warning("LHS has more args than RHS. RHS will be repeated.")
r <- extendToMatch(r, l)
}
for (II in 1:length(l)) {
do.call('<-', list(l[[II]], r[[II]]), envir=Envir)
}
}
# Used if LHS is larger than RHS
extendToMatch <- function(source, destin) {
s <- length(source)
d <- length(destin)
# Assume that destin is a length when it is a single number and source is not
if(d==1 && s>1 && !is.null(as.numeric(destin)))
d <- destin
dif <- d - s
if (dif > 0) {
source <- rep(source, ceiling(d/s))[1:d]
}
return (source)
}
# Grouping the left hand side
g = function(...) {
List = as.list(substitute(list(...)))[-1L]
class(List) = 'lbunch'
return(List)
}
Then to execute:
Group the left hand side using the new function g()
The right hand side should be a vector or a list
Use the newly-created binary operator %=%
# Example Call; Note the use of g() AND `%=%`
# Right-hand side can be a list or vector
g(a, b, c) %=% list("hello", 123, list("apples, oranges"))
g(d, e, f) %=% 101:103
# Results:
> a
[1] "hello"
> b
[1] 123
> c
[[1]]
[1] "apples, oranges"
> d
[1] 101
> e
[1] 102
> f
[1] 103
Example using lists of different sizes:
Longer Left Hand Side
g(x, y, z) %=% list("first", "second")
# Warning message:
# In `%=%.lbunch`(g(x, y, z), list("first", "second")) :
# LHS has more args than RHS. RHS will be repeated.
> x
[1] "first"
> y
[1] "second"
> z
[1] "first"
Longer Right Hand Side
g(j, k) %=% list("first", "second", "third")
# Warning message:
# In `%=%.lbunch`(g(j, k), list("first", "second", "third")) :
# RHS has more args than LHS. Only first2used.
> j
[1] "first"
> k
[1] "second"
Consider using functionality included in base R.
For instance, create a 1 row dataframe (say V) and initialize your variables in it. Now you can assign to multiple variables at once V[,c("a", "b")] <- values[c(2, 4)], call each one by name (V$a), or use many of them at the same time (values[c(5, 6)] <- V[,c("a", "b")]).
If you get lazy and don't want to go around calling variables from the dataframe, you could attach(V) (though I personally don't ever do it).
# Initialize values
values <- 1:100
# V for variables
V <- data.frame(a=NA, b=NA, c=NA, d=NA, e=NA)
# Assign elements from a vector
V[, c("a", "b", "e")] = values[c(2,4, 8)]
# Also other class
V[, "d"] <- "R"
# Use your variables
V$a
V$b
V$c # OOps, NA
V$d
V$e
here is my idea. Probably the syntax is quite simple:
`%tin%` <- function(x, y) {
mapply(assign, as.character(substitute(x)[-1]), y,
MoreArgs = list(envir = parent.frame()))
invisible()
}
c(a, b) %tin% c(1, 2)
gives like this:
> a
Error: object 'a' not found
> b
Error: object 'b' not found
> c(a, b) %tin% c(1, 2)
> a
[1] 1
> b
[1] 2
this is not well tested though.
A potentially dangerous (in as much as using assign is risky) option would be to Vectorize assign:
assignVec <- Vectorize("assign",c("x","value"))
#.GlobalEnv is probably not what one wants in general; see below.
assignVec(c('a','b'),c(0,4),envir = .GlobalEnv)
a b
0 4
> b
[1] 4
> a
[1] 0
Or I suppose you could vectorize it yourself manually with your own function using mapply that maybe uses a sensible default for the envir argument. For instance, Vectorize will return a function with the same environment properties of assign, which in this case is namespace:base, or you could just set envir = parent.env(environment(assignVec)).
As others explained, there doesn't seem to be anything built in. ...but you could design a vassign function as follows:
vassign <- function(..., values, envir=parent.frame()) {
vars <- as.character(substitute(...()))
values <- rep(values, length.out=length(vars))
for(i in seq_along(vars)) {
assign(vars[[i]], values[[i]], envir)
}
}
# Then test it
vals <- 11:14
vassign(aa,bb,cc,dd, values=vals)
cc # 13
One thing to consider though is how to handle the cases where you e.g. specify 3 variables and 5 values or the other way around. Here I simply repeat (or truncate) the values to be of the same length as the variables. Maybe a warning would be prudent. But it allows the following:
vassign(aa,bb,cc,dd, values=0)
cc # 0
list2env(setNames(as.list(rep(2,5)), letters[1:5]), .GlobalEnv)
Served my purpose, i.e., assigning five 2s into first five letters.
Had a similar problem recently and here was my try using purrr::walk2
purrr::walk2(letters,1:26,assign,envir =parent.frame())
https://stat.ethz.ch/R-manual/R-devel/library/base/html/list2env.html:
list2env(
list(
a=1,
b=2:4,
c=rpois(10,10),
d=gl(3,4,LETTERS[9:11])
),
envir=.GlobalEnv
)
If your only requirement is to have a single line of code, then how about:
> a<-values[2]; b<-values[4]
I'm afraid that elegent solution you are looking for (like c(a, b) = c(2, 4)) unfortunatelly does not exist. But don't give up, I'm not sure! The nearest solution I can think of is this one:
attach(data.frame(a = 2, b = 4))
or if you are bothered with warnings, switch them off:
attach(data.frame(a = 2, b = 4), warn = F)
But I suppose you're not satisfied with this solution, I wouldn't be either...
R> values = c(1,2,3,4)
R> a <- values[2]; b <- values[3]; c <- values[4]
R> a
[1] 2
R> b
[1] 3
R> c
[1] 4
Another version with recursion:
let <- function(..., env = parent.frame()) {
f <- function(x, ..., i = 1) {
if(is.null(substitute(...))){
if(length(x) == 1)
x <- rep(x, i - 1);
stopifnot(length(x) == i - 1)
return(x);
}
val <- f(..., i = i + 1);
assign(deparse(substitute(x)), val[[i]], env = env);
return(val)
}
f(...)
}
example:
> let(a, b, 4:10)
[1] 4 5 6 7 8 9 10
> a
[1] 4
> b
[1] 5
> let(c, d, e, f, c(4, 3, 2, 1))
[1] 4 3 2 1
> c
[1] 4
> f
[1] 1
My version:
let <- function(x, value) {
mapply(
assign,
as.character(substitute(x)[-1]),
value,
MoreArgs = list(envir = parent.frame()))
invisible()
}
example:
> let(c(x, y), 1:2 + 3)
> x
[1] 4
> y
[1]
Combining some of the answers given here + a little bit of salt, how about this solution:
assignVec <- Vectorize("assign", c("x", "value"))
`%<<-%` <- function(x, value) invisible(assignVec(x, value, envir = .GlobalEnv))
c("a", "b") %<<-% c(2, 4)
a
## [1] 2
b
## [1] 4
I used this to add the R section here: http://rosettacode.org/wiki/Sort_three_variables#R
Caveat: It only works for assigning global variables (like <<-). If there is a better, more general solution, pls. tell me in the comments.
For a named list, use
list2env(mylist, environment())
For instance:
mylist <- list(foo = 1, bar = 2)
list2env(mylist, environment())
will add foo = 1, bar = 2 to the current environement, and override any object with those names. This is equivalent to
mylist <- list(foo = 1, bar = 2)
foo <- mylist$foo
bar <- mylist$bar
This works in a function, too:
f <- function(mylist) {
list2env(mylist, environment())
foo * bar
}
mylist <- list(foo = 1, bar = 2)
f(mylist)
However, it is good practice to name the elements you want to include in the current environment, lest you override another object... and so write preferrably
list2env(mylist[c("foo", "bar")], environment())
Finally, if you want different names for the new imported objects, write:
list2env(`names<-`(mylist[c"foo", "bar"]), c("foo2", "bar2")), environment())
which is equivalent to
foo2 <- mylist$foo
bar2 <- mylist$bar
I want to assign multiple variables in a single line in R. Is it possible to do something like this?
values # initialize some vector of values
(a, b) = values[c(2,4)] # assign a and b to values at 2 and 4 indices of 'values'
Typically I want to assign about 5-6 variables in a single line, instead of having multiple lines. Is there an alternative?
I put together an R package zeallot to tackle this very problem. zeallot includes an operator (%<-%) for unpacking, multiple, and destructuring assignment. The LHS of the assignment expression is built using calls to c(). The RHS of the assignment expression may be any expression which returns or is a vector, list, nested list, data frame, character string, date object, or custom objects (assuming there is a destructure implementation).
Here is the initial question reworked using zeallot (latest version, 0.0.5).
library(zeallot)
values <- c(1, 2, 3, 4) # initialize a vector of values
c(a, b) %<-% values[c(2, 4)] # assign `a` and `b`
a
#[1] 2
b
#[1] 4
For more examples and information one can check out the package vignette.
There is a great answer on the Struggling Through Problems Blog
This is taken from there, with very minor modifications.
USING THE FOLLOWING THREE FUNCTIONS
(Plus one for allowing for lists of different sizes)
# Generic form
'%=%' = function(l, r, ...) UseMethod('%=%')
# Binary Operator
'%=%.lbunch' = function(l, r, ...) {
Envir = as.environment(-1)
if (length(r) > length(l))
warning("RHS has more args than LHS. Only first", length(l), "used.")
if (length(l) > length(r)) {
warning("LHS has more args than RHS. RHS will be repeated.")
r <- extendToMatch(r, l)
}
for (II in 1:length(l)) {
do.call('<-', list(l[[II]], r[[II]]), envir=Envir)
}
}
# Used if LHS is larger than RHS
extendToMatch <- function(source, destin) {
s <- length(source)
d <- length(destin)
# Assume that destin is a length when it is a single number and source is not
if(d==1 && s>1 && !is.null(as.numeric(destin)))
d <- destin
dif <- d - s
if (dif > 0) {
source <- rep(source, ceiling(d/s))[1:d]
}
return (source)
}
# Grouping the left hand side
g = function(...) {
List = as.list(substitute(list(...)))[-1L]
class(List) = 'lbunch'
return(List)
}
Then to execute:
Group the left hand side using the new function g()
The right hand side should be a vector or a list
Use the newly-created binary operator %=%
# Example Call; Note the use of g() AND `%=%`
# Right-hand side can be a list or vector
g(a, b, c) %=% list("hello", 123, list("apples, oranges"))
g(d, e, f) %=% 101:103
# Results:
> a
[1] "hello"
> b
[1] 123
> c
[[1]]
[1] "apples, oranges"
> d
[1] 101
> e
[1] 102
> f
[1] 103
Example using lists of different sizes:
Longer Left Hand Side
g(x, y, z) %=% list("first", "second")
# Warning message:
# In `%=%.lbunch`(g(x, y, z), list("first", "second")) :
# LHS has more args than RHS. RHS will be repeated.
> x
[1] "first"
> y
[1] "second"
> z
[1] "first"
Longer Right Hand Side
g(j, k) %=% list("first", "second", "third")
# Warning message:
# In `%=%.lbunch`(g(j, k), list("first", "second", "third")) :
# RHS has more args than LHS. Only first2used.
> j
[1] "first"
> k
[1] "second"
Consider using functionality included in base R.
For instance, create a 1 row dataframe (say V) and initialize your variables in it. Now you can assign to multiple variables at once V[,c("a", "b")] <- values[c(2, 4)], call each one by name (V$a), or use many of them at the same time (values[c(5, 6)] <- V[,c("a", "b")]).
If you get lazy and don't want to go around calling variables from the dataframe, you could attach(V) (though I personally don't ever do it).
# Initialize values
values <- 1:100
# V for variables
V <- data.frame(a=NA, b=NA, c=NA, d=NA, e=NA)
# Assign elements from a vector
V[, c("a", "b", "e")] = values[c(2,4, 8)]
# Also other class
V[, "d"] <- "R"
# Use your variables
V$a
V$b
V$c # OOps, NA
V$d
V$e
here is my idea. Probably the syntax is quite simple:
`%tin%` <- function(x, y) {
mapply(assign, as.character(substitute(x)[-1]), y,
MoreArgs = list(envir = parent.frame()))
invisible()
}
c(a, b) %tin% c(1, 2)
gives like this:
> a
Error: object 'a' not found
> b
Error: object 'b' not found
> c(a, b) %tin% c(1, 2)
> a
[1] 1
> b
[1] 2
this is not well tested though.
A potentially dangerous (in as much as using assign is risky) option would be to Vectorize assign:
assignVec <- Vectorize("assign",c("x","value"))
#.GlobalEnv is probably not what one wants in general; see below.
assignVec(c('a','b'),c(0,4),envir = .GlobalEnv)
a b
0 4
> b
[1] 4
> a
[1] 0
Or I suppose you could vectorize it yourself manually with your own function using mapply that maybe uses a sensible default for the envir argument. For instance, Vectorize will return a function with the same environment properties of assign, which in this case is namespace:base, or you could just set envir = parent.env(environment(assignVec)).
As others explained, there doesn't seem to be anything built in. ...but you could design a vassign function as follows:
vassign <- function(..., values, envir=parent.frame()) {
vars <- as.character(substitute(...()))
values <- rep(values, length.out=length(vars))
for(i in seq_along(vars)) {
assign(vars[[i]], values[[i]], envir)
}
}
# Then test it
vals <- 11:14
vassign(aa,bb,cc,dd, values=vals)
cc # 13
One thing to consider though is how to handle the cases where you e.g. specify 3 variables and 5 values or the other way around. Here I simply repeat (or truncate) the values to be of the same length as the variables. Maybe a warning would be prudent. But it allows the following:
vassign(aa,bb,cc,dd, values=0)
cc # 0
list2env(setNames(as.list(rep(2,5)), letters[1:5]), .GlobalEnv)
Served my purpose, i.e., assigning five 2s into first five letters.
Had a similar problem recently and here was my try using purrr::walk2
purrr::walk2(letters,1:26,assign,envir =parent.frame())
https://stat.ethz.ch/R-manual/R-devel/library/base/html/list2env.html:
list2env(
list(
a=1,
b=2:4,
c=rpois(10,10),
d=gl(3,4,LETTERS[9:11])
),
envir=.GlobalEnv
)
If your only requirement is to have a single line of code, then how about:
> a<-values[2]; b<-values[4]
I'm afraid that elegent solution you are looking for (like c(a, b) = c(2, 4)) unfortunatelly does not exist. But don't give up, I'm not sure! The nearest solution I can think of is this one:
attach(data.frame(a = 2, b = 4))
or if you are bothered with warnings, switch them off:
attach(data.frame(a = 2, b = 4), warn = F)
But I suppose you're not satisfied with this solution, I wouldn't be either...
R> values = c(1,2,3,4)
R> a <- values[2]; b <- values[3]; c <- values[4]
R> a
[1] 2
R> b
[1] 3
R> c
[1] 4
Another version with recursion:
let <- function(..., env = parent.frame()) {
f <- function(x, ..., i = 1) {
if(is.null(substitute(...))){
if(length(x) == 1)
x <- rep(x, i - 1);
stopifnot(length(x) == i - 1)
return(x);
}
val <- f(..., i = i + 1);
assign(deparse(substitute(x)), val[[i]], env = env);
return(val)
}
f(...)
}
example:
> let(a, b, 4:10)
[1] 4 5 6 7 8 9 10
> a
[1] 4
> b
[1] 5
> let(c, d, e, f, c(4, 3, 2, 1))
[1] 4 3 2 1
> c
[1] 4
> f
[1] 1
My version:
let <- function(x, value) {
mapply(
assign,
as.character(substitute(x)[-1]),
value,
MoreArgs = list(envir = parent.frame()))
invisible()
}
example:
> let(c(x, y), 1:2 + 3)
> x
[1] 4
> y
[1]
Combining some of the answers given here + a little bit of salt, how about this solution:
assignVec <- Vectorize("assign", c("x", "value"))
`%<<-%` <- function(x, value) invisible(assignVec(x, value, envir = .GlobalEnv))
c("a", "b") %<<-% c(2, 4)
a
## [1] 2
b
## [1] 4
I used this to add the R section here: http://rosettacode.org/wiki/Sort_three_variables#R
Caveat: It only works for assigning global variables (like <<-). If there is a better, more general solution, pls. tell me in the comments.
For a named list, use
list2env(mylist, environment())
For instance:
mylist <- list(foo = 1, bar = 2)
list2env(mylist, environment())
will add foo = 1, bar = 2 to the current environement, and override any object with those names. This is equivalent to
mylist <- list(foo = 1, bar = 2)
foo <- mylist$foo
bar <- mylist$bar
This works in a function, too:
f <- function(mylist) {
list2env(mylist, environment())
foo * bar
}
mylist <- list(foo = 1, bar = 2)
f(mylist)
However, it is good practice to name the elements you want to include in the current environment, lest you override another object... and so write preferrably
list2env(mylist[c("foo", "bar")], environment())
Finally, if you want different names for the new imported objects, write:
list2env(`names<-`(mylist[c"foo", "bar"]), c("foo2", "bar2")), environment())
which is equivalent to
foo2 <- mylist$foo
bar2 <- mylist$bar