Consider adding helper functions when creating composition data

[chantelwetzel-noaa]

Incorporating helper functions could streamline the code and reduce code redundancy within and across functions. Below are some example helper functions that could be considered:

# Two helper functions
make_long_comp <- function(data) {
  data |>
    dplyr::group_by(year, full_bin, sex) |>
    dplyr::summarize(
      expansion = sum(expansion)
    ) |>
    dplyr::group_by(year) |>
    dplyr::mutate(proportion = prop.table(expansion)) |>
    dplyr::select(-expansion)
}
pivot_wider_comp <- function(data) {
  data_long <- data |>
    tidyr::separate(full_bin, into = c("bin", "group")) |>
    dplyr::mutate(
      bin_factor = forcats::fct_cross(
        forcats::as_factor(as.numeric(bin)),
        forcats::as_factor(sex)
      )
    )
  temporary_levels <- levels(data_long[["bin_factor"]])
  data_wide <- data_long |>
    tidyr::pivot_wider(
      id_cols = year,
      names_from = bin_factor,
      values_from = proportion
    ) |>
    dplyr::select(year, temporary_levels)
  return(data_wide)
}
# Using the helper functions to create a list of length 3, basically one list with each of the output types for SS3, this would replace quite a bit of code that starts with the creation of `total_by_year`
attempt <- stratum_exp |>
  tidyr::pivot_longer(
    cols = c(-year, -strata, -bin),
    names_to = "sex",
    values_to = "expansion"
  ) |>
  dplyr::mutate(
    sex_group = ifelse(sex %in% c("female", "male"), "sexed", sex),
    full_bin = paste(bin, sex_group, sep = ":")
  ) |>
  dplyr::group_by(sex_group) |>
  dplyr::group_split() |>
  purrr::map(make_long_comp) |>
  purrr::map(pivot_wider_comp)

However, there are a few issues that would need to be sorted through prior to implementation:

1. Using an external vector in selections was deprecated in tidyselect 1.1.0. Please use all_of() or any_of() instead.
2. NaN and NA created need to be replaced with 0.
3. The composition data are not output in the order of the composition bin and would need to be reordered.