在 R 中使用混合类型将嵌套列表中的字符数字转换为数字

Question

I have a bunch of nested lists that arrive with all character-typed elements. Some of these characters are in fact numbers that need to be converted to numeric-typed elements.

How can I modify these lists so that all possible character-typed numbers are converted to numeric, but the non-number character elements remain intact (eg, aren't converted to NA by as.numeric() )?

For example, I start with

raw_list <- list(
      'a' = list('1', '2'),
      'b' = list(
        'c' = list('3', 'x'),
        'd' = c('4', 'z'),
        'e' = '5'))

But I need to get to

cleaned_list <- list(
      'a' = list(1, 2),
      'b' = list(
        'c' = list(3, 'x'),
        'd' = c(4, 'z'),
        'e' = 5))

Bonus: Extra gratitudie for tidyverse/purrr-based solutions

Answer 1

Another option is to use type.convert , which can take lists as arguments.

raw_list %>%
  type.convert(as.is = TRUE)

This would be the same as using purrr with type.convert .

library(purrr)

purrr::map(raw_list, type.convert, as.is = TRUE)

Output

$a
$a[[1]]
[1] 1

$a[[2]]
[1] 2


$b
$b$c
$b$c[[1]]
[1] 3

$b$c[[2]]
[1] "x"


$b$d
[1] "4" "z"

$b$e
[1] 5

Answer 2

You can use readr::parse_guess to convert character-typed numbers to numeric-typed numbers without converting characters to NA s.

I also created a function to convert the numbers in a nested list. Please note that you cannot store characters and numerics in the same vector.

I hope this would help you.

library(tidyverse)
func_parse_list <- function(l){
  if(length(l)>1){
    map(l, func_parse_list) %>% return()
  }else{
    parse_guess(l) %>% return()
  }
}
list('a' = list('1', '2'),
     'b' = list(
       'c' = list('3', 'x'),
       'd' = c('4', 'z'),
       'e' = '5')) %>%
  func_parse_list()
#> $a
#> $a[[1]]
#> [1] 1
#> 
#> $a[[2]]
#> [1] 2
#> 
#> 
#> $b
#> $b$c
#> $b$c[[1]]
#> [1] 3
#> 
#> $b$c[[2]]
#> [1] "x"
#> 
#> 
#> $b$d
#> $b$d[[1]]
#> [1] 4
#> 
#> $b$d[[2]]
#> [1] "z"
#> 
#> 
#> $b$e
#> [1] 5

^{Created on 2021-12-05 by the reprex package (v2.0.1)}