Convert number to alphabetically sortable character representation - r

Question

I need to pass some numbers into functions/environments/contexts that only accept strings. They will then be processed there, after receiving the result I can reconvert them to numbers. The key issue is that when the string representations are sorted, they must be sorted in the correct order (in what would be the numerical sort order of the numeric representation). I also need to be able to add numbers after creating the initial batch. Following are two conversion approaches and why they do not work:
The simplest conversion is the standard one. 1->"1", 10->"10" and so on. This does not satisfy the criteria of sortability, because "10" gets sorted before "2".
The next approach is to prefix with zeroes. 1->"001", 10->"010" and so on. This satisfies sortability, ("002" gets sorted before "010"), but if a a larger number needs to be added later, this approach fails. If, say the numbers 2000 and 10000 need to be added later it is not possible to do so in a way that maintains sorting.
Are there any good approaches to doing this? The question doen not pertain specifically to any particular language (although the target language in my use-case is R, which has a number of places such as vector names and others that accept only character variables). Simplicity and/or standardization (of the representation, not implementation-wise) would both be big factors in choosing the best solution here.

Answer 1

I've had the same problem I think, and I used workaround that could help you. But I'm not sure it will apply in all situations.
First, a vector with number as string ordered as string:
str.numbers <- sort(as.character(1:20))
Now, I used the numeric representation of the string to order numerically the same vector:
str.numbers[order(as.numeric(str.numbers))]
This does the trick for simple vectors. But not sure it'll solve more complex problems.

Answer 2

Funny how the world works, after searching for a solution but not finding it, I came across a package, with the last release only four days ago, that focuses on this problem. This is the strex package. A reproducible example of my troubles, and how strex provides a pretty good fix, follows:
# Load the strex library
library(strex)
#> Loading required package: stringr
# This would be a function created by others, doing more than
# sorting, but nevertheless requiring sortable input
fun_requiring_char <- function(x) {
stopifnot(is.character(x))
sort(x)
}
# Example data
set.seed(42)
a <- sample(20, 5) ; a
#> [1] 17 5 1 10 4
b <- sample(2000, 10) ; b
#> [1] 1170 634 49 1152 1327 24 1863 356 1625 165
# Won't work, error
#fun_requiring_char(a)
# Works, but returns incorrectly sorted input
fun_requiring_char(as.character(a))
#> [1] "1" "10" "17" "4" "5"
fun_requiring_char(as.character(b))
#> [1] "1152" "1170" "1327" "1625" "165" "1863" "24" "356" "49" "634"
# Solution provided by strex
fun_requiring_char(str_alphord_nums(a))
#> [1] "01" "04" "05" "10" "17"
fun_requiring_char(str_alphord_nums(b))
#> [1] "0024" "0049" "0165" "0356" "0634" "1152" "1170" "1327" "1625" "1863"
# What quick and dirty zero padding did not allow was to first
# convert a, and then b into character, where both were of a
# unified format that would represent the numbers and yet be
# sortable in correct order according to the numbers they
# represent. However, using str_alphord_nums repeatedly gets
# very close to a solution.
ac <- str_alphord_nums(a); ac
#> [1] "17" "05" "01" "10" "04"
bc <- str_alphord_nums(b); bc
#> [1] "1170" "0634" "0049" "1152" "1327" "0024" "1863" "0356" "1625" "0165"
# Wrong order here
fun_requiring_char(c(ac,bc))
#> [1] "0024" "0049" "01" "0165" "0356" "04" "05" "0634" "10" "1152"
#> [11] "1170" "1327" "1625" "17" "1863"
# But doing an alphord again on the concatenated vectors provides a fix
fun_requiring_char(str_alphord_nums(c(ac,bc)))
#> [1] "0001" "0004" "0005" "0010" "0017" "0024" "0049" "0165" "0356" "0634"
#> [11] "1152" "1170" "1327" "1625" "1863"
Created on 2020-10-21 by the reprex package (v0.3.0)