Unlist the second to last list of a nested list-CodePudding

I have a deeply nested list of lists. In the "center" of the nested list is a vector containing n integers. I need to count how many integers are in each nested list, then unlist one level above to have a vector of these counts (i.e., instead of list(0, 1:5, 0, 0, 1:3) at the center of the nest, I want c(0, 5, 0, 0, 3).

This seems relatively simple - I was able to use rapply to accomplish the first part, i.e. convert list(0, 1:5, 0, 0, 1:3) to list(0, 5, 0, 0, 3). My specific question I need help with is how to unlist the innermost lists to a vector (instead of list(0, 5, 0, 0, 3) I want c(0, 5, 0, 0, 3)

I have searched and tried various apply, lapply, unlist approaches but none of them are quite right as they target the very innermost list. Since the list I want to unlist is the second to last element, I am struggling finding a way to accomplish this elegantly.

In the sample data below, I can get the desired outcome 2 ways: either multiple lapply functions or a for loop. However, my actual data contain many more lists and millions of datapoints, so these are likely not effective options.

Below is (1) sample data, (2) what I have tried, and (3) sample data having the desired structure.

Sample Data

have_list <- list(scenario1 = list(method1 = list(place1 = list(0, 1:5, 0, 0, 1:3),
                                                  place2 = list(1:2, 0, 1:10, 0, 0),
                                                  place3 = list(0:19, 0, 0, 0, 0),
                                                  place4 = list(1:100, 0, 0, 1:4, 0)),
                                   method2 = list(place1 = list(1:5, 1:5, 0, 0, 1:3),
                                                  place2 = list(0, 0, 1:5, 0, 0),
                                                  place3 = list(0:19, 0, 1:7, 0, 0),
                                                  place4 = list(1:22, 0, 0, 1:4, 0)),
                                   method3 = list(place1 = list(0, 1:2, 1:6, 0, 1:3),
                                                  place2 = list(1:2, 0, 1:6, 1:4, 0),
                                                  place3 = list(0:19, 0, 0, 0, 1:2),
                                                  place4 = list(1:12, 0, 0, 1:12, 0))),
                  scenario2 = list(method1 = list(place1 = list(0, 1:5, 0, 0, 1:3),
                                                  place2 = list(1:2, 0, 1:10, 0, 0),
                                                  place3 = list(0:19, 0, 0, 0, 0),
                                                  place4 = list(1:100, 0, 0, 1:4, 0)),
                                   method2 = list(place1 = list(1:5, 1:5, 0, 0, 1:3),
                                                  place2 = list(0, 0, 1:5, 0, 0),
                                                  place3 = list(0:19, 0, 1:7, 0, 0),
                                                  place4 = list(1:22, 0, 0, 1:4, 0)),
                                   method3 = list(place1 = list(0, 1:2, 1:6, 0, 1:3),
                                                  place2 = list(1:2, 0, 1:6, 1:4, 0),
                                                  place3 = list(0:19, 0, 0, 0, 1:2),
                                                  place4 = list(1:12, 0, 0, 1:12, 0))))

What I have tried

And SO questions I have visited:

# Get number of integers in each nested list 
lengths <- rapply(have_list, function(x) unlist(length(x)), how = "list") # this works fine

#' Each count is currently still in its own list of length 1,
#' Convert each count to vector
#' In the "middle" the nested list:
    # I have list(0, 5, 0, 0, 3) 
    # I want c(0, 5, 0, 0, 3)

# Attempts to unlist the counts
# Unlist the counts
test1 <- rapply(lengths, unlist, how = "list") # doesn't work
test2 <- unlist(lengths, recursive = FALSE) # doesn't work
test3 <- lapply(lengths, function(x) lapply(x, unlist)) # doesnt work
test4 <- lapply(lengths, function(x) lapply(x, unlist, recursive = FALSE)) # doesnt work 
test5 <- rapply(have_list, function(x) unlist(length(x)), how = "list")  #doesnt work
test6 <- rapply(have_list, function(x) unlist(length(x)), how = "unlist")  #doesnt work

Data structure I want

# This works on test data but is impractical for real data
want_list <- lapply(lengths, function(w) lapply(w, function(x) lapply(x, unlist)))

# or

want_list <- lengths 

## for loops work but is not practical

for (i in 1:length(lengths)){
  for (j in 1:length(lengths[[i]])){
    for (k in 1:length(lengths[[i]][[j]])){
      want_list[[i]][[j]][[k]] <- unlist(lengths[[i]][[j]][[k]])
    }
  }
}

CodePudding user response：

An option is to melt the nested list with rrapply, replace the 'value' column with the lengths and then use the recursive split (rsplit) from collapse

library(rrapply)
library(collapse)
dat <- transform(rrapply(have_list, how = "melt"), value= lengths(value))
out <- rsplit(dat$value, dat[1:3])

-testing with OP' expected

identical(out, want_list)
[1] TRUE

CodePudding user response：

This can be done by using recursion. A simple recursion will be:

my_fun <- function(x) if(is.list(x[[1]])) lapply(x, my_fun) else lengths(x)

out <- my_fun(have_list)

identical(out, want_list)
[1] TRUE