Sorting objects by size with unit denominations in a vector or data frame in R-CodePudding

I am trying to sort a vector of object sizes in descending order and create a dataframe. I came across an issue with the sorting because the numbers have unit denominations (e.g. Kb, Mb, etc.), and I was wondering how I can sort numbers in ascending or descending order? Because the number have denominations they are essentially treated as character vectors, and can therefore not be sorted after size.

Example 1:

library(dplyr)

l <- list(1:1e6, 1:1e1, 1:1e3, 1:1e7)
l <- sapply(
    l,
    function(x){
        object.size(x) %>% format(units = "auto")
    }
)

# Alt. A: Sorting the vector before coercing to dataframe
sort(l) %>% as.data.frame() 

A data.frame: 4 × 1
.
<chr>
96 bytes
4 Kb
38.1 Mb
3.8 Mb

# Alt. B: Coerce to dataframe then sort using arrange()
as.data.frame(l) %>% arrange(desc(names(.)[1]))

A data.frame: 4 × 1
l
<chr>
3.8 Mb
96 bytes
4 Kb
38.1 Mb

Desired output:

A data.frame: 4 × 1
l
<chr>
38.1 Mb
3.8 Mb
4 Kb
96 bytes

CodePudding user response：

The problem is that your sapply loop only keeps the formatted output, which is much harder to sort. Using purrr you can store two values for each iteration in a data frame and bind the results together. So you can do:

library(dplyr)

l <- list(1:1e6, 1:1e1, 1:1e3, 1:1e7)
l_1 <- purrr::map_df(l, function(x) {
  tibble(
    size_raw = object.size(x),
    size = size_raw %>% format(units = "auto")
  )
})

l_1 %>% 
  arrange(-size_raw)
#> # A tibble: 4 × 2
#>   size_raw       size    
#>   <objct_sz>     <chr>   
#> 1 40000048 bytes 38.1 Mb 
#> 2 4000048 bytes  3.8 Mb  
#> 3 4048 bytes     4 Kb    
#> 4 96 bytes       96 bytes

^{Created on 2022-03-16 by the reprex package (v2.0.1)}