CODEWITHSUNDEEP
renvironment-variablesenvironmentassignment-operator
Robbin
Robbin

Reputation: 33

Assign values to part of an object in another environment in R

Suppose I have two similar big objects x, y (datatable) defined in an envirnomnet e. I would like to change a big part of x or y in a similar way using a function f without creating a copy of x or y within the executive environment of f. Example:

e <- new.env()
e$x <- c(1,2,3) # imagine this to be BIG (ie. dataframe with 200k vars each 500k rows)
e$y <- c(4,5,6) # same here
e$v <- 2        # minor variables 

f <- function(var_str, env, input){

    # do some computation on parts of var_str which is either "x" or "y"
    # and store these right back into e$x or e$y, respectively.
    # ie

    str <- paste0(var_str,"[2:3] <- (",var_str,"[2:3])^2 + rep(v,2) + ", deparse(input1),"^3/c(100,101)")
    eval(parse(text=str), envir= e)

    # this does work but I can image there is an easier/more elegant way 
    # of doing this.
}

I would like to define the function in the global environment and apply this function to e$x and e$y with different variables in input. Ie. execute

f("x", e, c(1,2))
f("y", e, c(3,4))

Does anybody have an elegant solution to this.

Upvotes: 2

Views: 1315

Answers (2)

IRTFM
IRTFM

Reputation: 263332

Out of curiosity I tried working with the simple example in the R6 tutorial and came up with this (copied as a console transcript). I honestly didn't know if it complies with the non-copy requirements of the request, but it does appear to modify an object in-place.

# Assumes one has created an R6 constructor named `RC` ...
# slightly extended from the example in section
RC <- setRefClass("RC",
  fields = list(x = 'ANY'),
  methods = list(
    getx = function() x,
    setx = function(value) x <<- value,
    setsub = function(i,j,val) x[i,j] <<- val
  )
)

#--- execution --- 
rc <- RC$new()
rc$setx(matrix(1:20, 4,5))
rc
# --- result ---    
Reference class object of class "RC"

Field "x":
     [,1] [,2] [,3] [,4] [,5]
[1,]    1    5    9   13   17
[2,]    2    6   10   14   18
[3,]    3    7   11   15   19
[4,]    4    8   12   16   20

Test the setsub class-specific function:

rc$setsub(4,5,0)   #Test of setting a single element within an object to a new value
rc
#-------    
Reference class object of class "RC"
Field "x":
     [,1] [,2] [,3] [,4] [,5]
[1,]    1    5    9   13   17
[2,]    2    6   10   14   18
[3,]    3    7   11   15   19
[4,]    4    8   12   16    0

rc$setsub(1:4,1:4,0)    #Test of setting a range of elements within an object to a new value
rc
# ---------
Reference class object of class "RC"
Field "x":
     [,1] [,2] [,3] [,4] [,5]
[1,]    0    0    0    0   17
[2,]    0    0    0    0   18
[3,]    0    0    0    0   19
[4,]    0    0    0    0    0

So this (very lightly tested) implementation of the [<--function via setsub does succeed. A single effort to extend this to the example using a self reference did fail.

Upvotes: 0

Roland
Roland

Reputation: 132706

eval(parse()) is to be avoided. You can reference the environment like this:

e <- new.env()
e$x <- c(1,2,3) # imagine this to be BIG (ie. dataframe with 200k vars each 500k rows)
e$y <- c(4,5,6) # same here
e$v <- 2        # minor variables 

f <- function(var_str, env, input){

  # do some computation on parts of var_str which is either "x" or "y"
  # and store these right back into e$x or e$y, respectively.
  # ie
  env[[var_str]][2:3] <- (env[[var_str]][2:3])^2 + rep(env$v,2) + input^3/input

}


f("x", e, 1:2)
e$x
#[1]  1  7 15

Upvotes: 3

Related Questions