feat: finalize_characters()

[jonthegeek]

Summary

As a data wrangler, in order to work with categorical data in the most appropriate type, I would like to automatically convert character columns to factors when they have a small number of unique values.

Proposed signature

finalize_characters(.dataset, max_unique = 30L)

Arguments

• .dataset (data.frame) — The dataset to process.
• max_unique (integer(1)) — Maximum number of distinct values a character column may have to be eligible for conversion. Defaults to 30L.

Returns the .dataset with qualifying character columns converted to factor.

Behavior

• Iterates over all columns where is.character(col) is TRUE.
• For each such column, counts the number of distinct non-NA values.
• If that count is <= max_unique, converts the column to factor using factor(col).
• Columns that are not character, or that exceed max_unique distinct values, are left unchanged.
• NA values in a character column are preserved as NA in the resulting factor.

Details

We already import {stbl}, so use stbl::to_fct() for the factor casting.

This function should "match" finalize_integers(). If anything is repeated, abstract the shared code and update finalize_integers(). For example, the dataset check at the top of finalize_integers() should be repeated here, and thus abstracted to a .check_dataset_is_listish() helper defined in aaa-conditions.R. Be sure to accept a call argument in .check_dataset_is_listish() (default = rlang::caller_env()), and pass it on to .pkg_abort().