R/textfeature.R

In tidymodels/textrecipes: Extra 'Recipes' for Text Processing

#' Calculate Set of Text Features
#'
#' `step_textfeature()` creates a *specification* of a recipe step that will
#' extract a number of numeric features of a text column.
#'
#' @template args-recipe
#' @template args-dots
#' @template args-role_predictors
#' @template args-trained
#' @template args-columns
#' @param extract_functions A named list of feature extracting functions.
#'   Defaults to `count_functions`. See details for more information.
#' @param prefix A prefix for generated column names, defaults to "textfeature".
#' @template args-keep_original_cols
#' @template args-skip
#' @template args-id
#'
#' @template returns
#'
#' @details
#'
#' This step will take a character column and returns a number of numeric
#' columns equal to the number of functions in the list passed to the
#' `extract_functions` argument.
#'
#' All the functions passed to `extract_functions` must take a character vector
#' as input and return a numeric vector of the same length, otherwise an error
#' will be thrown.
#'
#' # Tidying
#'
#' When you [`tidy()`][recipes::tidy.recipe()] this step, a tibble is returned with
#' columns `terms`, `functions`, and `id`:
#'
#' \describe{
#'   \item{terms}{character, the selectors or variables selected}
#'   \item{functions}{character, name of feature functions}
#'   \item{id}{character, id of this step}
#' }
#'
#' @template case-weights-not-supported
#'
#' @family Steps for Numeric Variables From Characters
#'
#' @examplesIf rlang::is_installed("modeldata")
#' library(recipes)
#' library(modeldata)
#' data(tate_text)
#'
#' tate_rec <- recipe(~., data = tate_text) |>
#'   step_textfeature(medium)
#'
#' tate_obj <- tate_rec |>
#'   prep()
#'
#' bake(tate_obj, new_data = NULL) |>
#'   slice(1:2)
#'
#' bake(tate_obj, new_data = NULL) |>
#'   pull(textfeature_medium_n_words)
#'
#' tidy(tate_rec, number = 1)
#' tidy(tate_obj, number = 1)
#'
#' # Using custom extraction functions
#' nchar_round_10 <- function(x) round(nchar(x) / 10) * 10
#'
#' recipe(~., data = tate_text) |>
#'   step_textfeature(medium,
#'     extract_functions = list(nchar10 = nchar_round_10)
#'   ) |>
#'   prep() |>
#'   bake(new_data = NULL)
#' @export
step_textfeature <-
  function(
    recipe,
    ...,
    role = "predictor",
    trained = FALSE,
    columns = NULL,
    extract_functions = count_functions,
    prefix = "textfeature",
    keep_original_cols = FALSE,
    skip = FALSE,
    id = rand_id("textfeature")
  ) {
    recipes::recipes_pkg_check(required_pkgs.step_textfeature())

    add_step(
      recipe,
      step_textfeature_new(
        terms = enquos(...),
        role = role,
        trained = trained,
        columns = columns,
        extract_functions = extract_functions,
        prefix = prefix,
        keep_original_cols = keep_original_cols,
        skip = skip,
        id = id
      )
    )
  }

step_textfeature_new <-
  function(
    terms,
    role,
    trained,
    columns,
    extract_functions,
    prefix,
    keep_original_cols,
    skip,
    id
  ) {
    step(
      subclass = "textfeature",
      terms = terms,
      role = role,
      trained = trained,
      columns = columns,
      extract_functions = extract_functions,
      prefix = prefix,
      keep_original_cols = keep_original_cols,
      skip = skip,
      id = id
    )
  }

#' @export
prep.step_textfeature <- function(x, training, info = NULL, ...) {
  col_names <- recipes_eval_select(x$terms, training, info)

  check_string(x$prefix, arg = "prefix")

  training <- factor_to_text(training, col_names)

  check_type(training[, col_names], types = c("string", "factor", "ordered"))

  purrr::walk(x$extract_functions, validate_string2num)

  step_textfeature_new(
    terms = x$terms,
    role = x$role,
    trained = TRUE,
    columns = col_names,
    extract_functions = x$extract_functions,
    prefix = x$prefix,
    keep_original_cols = get_keep_original_cols(x),
    skip = x$skip,
    id = x$id
  )
}

#' @export
bake.step_textfeature <- function(object, new_data, ...) {
  col_names <- object$columns
  check_new_data(col_names, object, new_data)

  new_data <- factor_to_text(new_data, col_names)

  for (col_name in col_names) {
    tf_text <- map_dfc(object$extract_functions, \(f) f(new_data[[col_name]]))

    colnames(tf_text) <- paste(
      object$prefix,
      col_name,
      colnames(tf_text),
      sep = "_"
    )

    tf_text <- recipes::check_name(tf_text, new_data, object, names(tf_text))

    new_data <- vec_cbind(new_data, tf_text)
  }

  new_data <- remove_original_cols(new_data, object, col_names)

  new_data
}

#' @export
print.step_textfeature <-
  function(x, width = max(20, options()$width - 30), ...) {
    title <- "Text feature extraction for "
    print_step(x$columns, x$terms, x$trained, title, width)
    invisible(x)
  }

#' @rdname step_textfeature
#' @usage NULL
#' @export
tidy.step_textfeature <- function(x, ...) {
  if (is_trained(x)) {
    if (length(x$columns) == 0) {
      res <- tibble(
        terms = character(),
        functions = character()
      )
    } else {
      res <- tibble(
        terms = rep(unname(x$columns), each = length(x$extract_functions)),
        functions = rep(names(x$extract_functions), length(x$terms))
      )
    }
  } else {
    term_names <- sel2char(x$terms)
    res <- tibble(
      terms = term_names,
      functions = NA_character_
    )
  }
  res$id <- x$id
  res
}

validate_string2num <- function(fun) {
  string <- c("This is a test string", "with", "3 elements")

  out <- fun(string)
  if (!(is.numeric(out) | is.logical(out))) {
    cli::cli_abort("Function {.fn {fun}} must return a numeric.")
  }

  if (length(string) != length(out)) {
    cli::cli_abort(
      "{.fn {deparse(substitute(fun))}} must return the same length output as 
      its input."
    )
  }
}

#' @rdname required_pkgs.step
#' @export
required_pkgs.step_textfeature <- function(x, ...) {
  c("textrecipes")
}