Update apply_fmt_fn() to not overwrite existing formatted statistics

[ddsjoberg]

In some cases, users want apply formatting functions across a statistic. in the example below, the rounding function ensures the rounded statistics sum to the same number as the raw stats.

Perhaps we could also introduce a apply_fmt_fn(by) argument, and the statistics would be applied to the group of statistics. My only reservation is that this would fail when there are a mix a formatting functions that get applied to vectors and one-in-one-out formatting function (which would apply to more complex cases). But in those cases, the users can easily program around it.

library(cards)

ex <-
  ard_categorical(mtcars, variables = cyl, statistic = ~"p") |> 
  dplyr::mutate(
    .by = "stat_name",
    stat_fmt = 
      if (stat_name[1] == "p") JLutils::round_preserve_sum(unlist(stat) * 100, digits = 2) |> as.list()
      else purrr::map(stat, ~NULL)
  )
#> Registered S3 method overwritten by 'GGally':
#>   method from   
#>   +.gg   ggplot2
ex
#> {cards} data frame: 3 x 10
#>   variable variable_level stat_name stat_label  stat stat_fmt
#> 1      cyl              4         p          % 0.344    34.37
#> 2      cyl              6         p          % 0.219    21.88
#> 3      cyl              8         p          % 0.438    43.75
#> ℹ 4 more variables: context, fmt_fn, warning, error

ex |> apply_fmt_fn()
#> {cards} data frame: 3 x 10
#>   variable variable_level stat_name stat_label  stat stat_fmt
#> 1      cyl              4         p          % 0.344     34.4
#> 2      cyl              6         p          % 0.219     21.9
#> 3      cyl              8         p          % 0.438     43.8
#> ℹ 4 more variables: context, fmt_fn, warning, error

^{Created on 2024-08-02 with reprex v2.1.0}