Determine level of nesting in R?

Question

Is there an easy way (i.e. a function) to determine the level of nesting in list? I know there is str which can be used to get this information. But is there something that simply gives back the result? And can I use such a function to get the names of all levels of alist (recursively) ?

You could try `s <- (capture.output(str(mylist, nest.lev=1))[-1])` and then use string processing to capture the number of `..`s that start each element of `s`. (If I get 10 minutes later and there's no better answer by then, I'll give that a shot myself.) — Josh O'Brien, Nov 17 '12 at 18:13

score 26 · Accepted Answer · answered Nov 17 '12 at 18:52

26

A little recursive function can do this for you:

depth <- function(this,thisdepth=0){
  if(!is.list(this)){
    return(thisdepth)
  }else{
    return(max(unlist(lapply(this,depth,thisdepth=thisdepth+1))))    
  }
}

If you've got package:testthat, here's a test set:

l1=list(1,2,3)
l2=list(1,2,l1,4)
l3=list(1,l1,l2,5)

require(testthat)
expect_equal(depth(l1),1)
expect_equal(depth(l2),2)
expect_equal(depth(l3),3)

Apologies for using lower-case L in variable names. Readability fail.

answered Nov 17 '12 at 18:52

Spacedman

92,590
12
140
224

17

+1 -- A bit shorter would be `depth <- function(this) ifelse(is.list(this), 1L + max(sapply(this, depth)), 0L)` – flodel Nov 17 '12 at 19:59
1

Nice improvement! Passes my tests. R Code Golf anyone? – Spacedman Nov 18 '12 at 00:39
Thanks guys. Very nice indeed. I am somewhat surprised that there is no base solution for this. – Matt Bannert Nov 19 '12 at 08:55
2

list() returns -inf, quick fix: if(is.list(this) && length(this) == 0){return(0)} – kevin Feb 13 '19 at 13:45

score 7 · Answer 2 · answered Apr 24 '17 at 15:58

7

You can now use depth() from the purrr package!

Note: currently the function is part of the development version of purrr but will become part of the official CRAN version once the package gets a version bump

answered Apr 24 '17 at 15:58

Manuel R

3,976
4
28
41

8

I think its now in CRAN as `purrr::vec_depth`, but there's also `plotrix::listDepth` – Spacedman Mar 13 '18 at 10:36
`listDepth` was removed from the `plotrix` package in version 3.8-1. But +1 for `purrr::vec_depth`, which gets the job done. – dbc Sep 23 '21 at 15:14

score 2 · Answer 3 · answered Nov 17 '12 at 17:37

2

If all elements are named, you could use this (from the code of unlist):

mylist <- list(a=list(x=1),b=list(c=list(y=c(2,3)),d=c("a","b")))
names(.Internal(unlist(mylist, TRUE, TRUE)))
#[1] "a.x"    "b.c.y1" "b.c.y2" "b.d1"   "b.d2"

answered Nov 17 '12 at 17:37

Roland

127,288
10
191
288

score 1 · Answer 4 · answered Jan 27 '21 at 12:23

Another approach using rrapply() in the rrapply-package (extension of base rapply()):

library(rrapply)

l1 <- list(1, 2, 3)
l2 <- list(1, 2, l1, 4)
l3 <- list(1, l1, l2, 5)

max(rrapply(l1, f = function(x, .xpos) length(.xpos), how = "unlist"))
#> [1] 1
max(rrapply(l2, f = function(x, .xpos) length(.xpos), how = "unlist"))
#> [1] 2
max(rrapply(l3, f = function(x, .xpos) length(.xpos), how = "unlist"))
#> [1] 3

Determine level of nesting in R?

4 Answers4

Linked