在R中处理字符串中的数字

Question

I have the following (numbers-containing) character vector:

nums = c("1, 2", "1, 2, 4", "2, 4", "1, 2, 3, 4, 5", "2, 3, 5", NA, NA, NA, NA)

I want to set an algorithm that test if n subset of elements within nums contain n unique numbers and then remove those numbers from other elements. Where n is any number from 1 to 9 .

In the example above, as the first 3 elements contain only 3 numbers: 1, 2, 4 , these numbers should be removed from other elements. So the output would be like:

nums = c("1, 2", "1, 2, 4", "2, 4", "3, 5", "3, 5", NA, NA, NA, NA)

Note that it could be 2 elements having 2 unique numbers or 4 elements having 4 unique numbers, ... etc.

I'd like to keep the final output as a character vector of the same length as the original.

Answer 1

If I understand well, u can apply something like the following:

library(stringr)
library(readr)
library(purrr)
nums = c("1, 2", "1, 2, 4", "2, 4", "1, 2, 3, 4, 5", "2, 3, 5", NA, NA, NA, NA)

# create a list within each element is a character element of nums
num_into_list <- stringr::str_split(nums, ",")

# convert to numbers
num_into_list <- purrr::map(num_into_list, readr::parse_number)

# collect unique numbers from the nth first subset of the list (example 3)
not_allowed <- unique(unlist(num_into_list[1:3]))

# filter only values on the rest of the subset that doesn't contain
# values in not_allowed vector, using a logical subsetting operation
# inside of  anonymous function (purrr shortcut to create this)   
output_list <- c(num_into_list[1:3],   # first 3 subset are the same
                 purrr::map(num_into_list[4:9], ~ .[!(. %in% not_allowed)]))

# finally convert into a chr vector
output <- unlist(output_list)

You can make a function with the above code if parametrize the nth first subset to create the not_allowed vector and the length of your vector to then reconstitute the list (in the output_list step indexation).