Capturing expressions inside lists as text

Question

I currently have this function, which takes a table and two lists of expressions, and evaluates them, turning them into two matrices. I use two lists instead of ... because I need to be able to determine whether the expressions are going to fall in Y or Z.

func = function(tb, Y, Z) {
  Y_matrix = enquo(Y) %>% 
    eval_tidy(tb) %>%
    as.data.frame %>%
    as.matrix

  Z_matrix = enquo(Z) %>%
    eval_tidy(tb) %>%
    as.data.frame %>%
    as.matrix

  list(Y_matrix, Z_matrix)
}

For instance:

> tb = data.frame(a = 1:3, b = 3:1, c = 2*(1:3))
> func(tb, list(a + b, a - b), list(a*c, a + b + c))
[[1]]
     c.4L..4L..4L. c..2L..0L..2L.
[1,]             4             -2
[2,]             4              0
[3,]             4              2

[[2]]
     c.2..8..18. c.6..8..10.
[1,]           2           6
[2,]           8           8
[3,]          18          10

However, I also want to capture the expressions inside the lists as text, so I could use them to name the columns. For instance, my desired output would be something like:

> func(tb, list(a + b, a - b), list(a*c, a + b + c))
[[1]]
             a+b            a-b
[1,]             4             -2
[2,]             4              0
[3,]             4              2

[[2]]
              a*c         a + b + c
[1,]           2           6
[2,]           8           8
[3,]          18          10

How can I do this?

score 3 · Accepted Answer · answered May 10 '18 at 19:12

You can use something like deparse to go over the language elements in your list to turn those into strings. For example

func = function(tb, Y, Z) {
  Y_names <- map_chr(as.list(enexpr(Y))[-1], deparse)
  Y_matrix = enquo(Y) %>% 
    eval_tidy(tb) %>%
    as.data.frame %>%
    set_names(Y_names) %>% 
    as.matrix

  Z_names <- map_chr(as.list(enexpr(Z))[-1], deparse)
  Z_matrix = enquo(Z) %>%
    eval_tidy(tb) %>%
    as.data.frame %>%
    set_names(Z_names) %>% 
    as.matrix

  list(Y_matrix, Z_matrix)
}

which works with this example

func(tb, list(a + b, a - b), list(a*c, a + b + c))
# [[1]]
#      a + b a - b
# [1,]     4    -2
# [2,]     4     0
# [3,]     4     2

# [[2]]
#      a * c a + b + c
# [1,]     2         6
# [2,]     8         8
# [3,]    18        10

Wow! Thanks! Wasn't really aware that `enexpr()` objects had an easily accessible tree structure. Now I feel like doing more advanced stuff. Thanks! — Luiz Rodrigo, May 11 '18 at 02:37

score 2 · Answer 2 · answered May 10 '18 at 19:08

2

We could pass the arguments as a quosure and then do the evaluation (!!!)

library(tidyverse)
f1 <- function(tb, Y, Z) {
       Y_matrix <-  tb %>% 
             transmute(!!! Y)

       Z_matrix <- tb %>%
             transmute(!!! Z)

       list(Y_matrix, Z_matrix)


   }

f1(tb, quos(a + b, a - b), quos(a*c, a + b + c))
#[[1]]
#  a + b a - b
#1     4    -2
#2     4     0
#3     4     2

#[[2]]
#  a * c a + b + c
#1     2         6
#2     8         8
#3    18        10

answered May 10 '18 at 19:08

akrun

874,273
37
540
662

Nice workaround! I'm not choosing it as my favourite answer only because it has the inconvenience of forcing the user of the function to have rlang loaded. – Luiz Rodrigo May 11 '18 at 02:23

Capturing expressions inside lists as text

2 Answers2