parse_factor implicitly relevels empty string "" and actual NA to factor "NA"

[djbirke]

readr_1.1.1 and tidyverse_1.2.1

I want to convert a string column to factors, taking "NA" as an actual factor level (say "NA" means "North America"), and also taking "" as an actual factor level. However, both "" and <NA> are mapped to "NA" by parse_factor, while NC is mapped to <NA>.

library(tidyverse)

df <- tibble(
  id = 1:5,
  group = c('NA', 'NB', 'NC', '', NA)
)

df %>%
  mutate(f = parse_factor(group,
                          levels = c('', 'NA', 'NB'),
                          na = character()))
#> Warning: 1 parsing failure.
#> row # A tibble: 1 x 4 col     row   col expected           actual expected   <int> <int> <chr>              <chr>  actual 1     3    NA value in level set NC
#> # A tibble: 5 x 3
#>      id group f    
#>   <int> <chr> <fct>
#> 1     1 NA    NA   
#> 2     2 NB    NB   
#> 3     3 NC    <NA> 
#> 4     4 ""    NA   
#> 5     5 <NA>  NA

But I expected

#> # A tibble: 5 x 3
#>      id group f    
#>   <int> <chr> <fct>
#> 1     1 NA    NA   
#> 2     2 NB    NB   
#> 3     3 NC    <NA> 
#> 4     4 ""    ""
#> 5     5 <NA>  <NA>

[jimhester]

Thanks, should now be fixed!

library(readr)
x <- c("", "NC", "NC", "NC", "", "", "NB", "NA")

tibble::as_tibble(parse_factor(x, levels = c("NA", "NB", "NC", ""), na = character()))
#> # A tibble: 8 x 1
#>   value
#>   <fct>
#> 1 ""   
#> 2 NC   
#> 3 NC   
#> 4 NC   
#> 5 ""   
#> 6 ""   
#> 7 NB   
#> 8 NA

^{Created on 2018-11-14 by the reprex package (v0.2.1)}

[djbirke]

Awesome, thank you for your work!