# Join utilities

#' Test the join between two sets of identifiers
#'
#' @param x A vector of identifiers to check against `y`.
#' @param y A vector of identifiers to look for a match in.
#' @param distinct Logical. Should duplicate values of `x` and `y` be removed
#'   before testing? Default is `TRUE`.
#'
#' @return Invisibly returns `NULL`; prints a formatted summary of match
#'   statistics to the console via [writeLines()].
#' @export
#'
#' @examples
#' x <- LETTERS
#' y <- c(letters, LETTERS)
#' match_test(x, y)
match_test <- function(x, y, distinct = TRUE) {
  if (distinct) {
    x <- unique(x)
    y <- unique(y)
  }

  # TODO: DO not report 100% if there is even 1 mismatch

  xiny     <- sum(x %in% y)
  total_x  <- length(x)
  pct_x    <- round(100 * xiny / total_x, 2)

  yinx     <- sum(y %in% x)
  total_y  <- length(y)
  pct_y    <- round(100 * yinx / total_y, 2)

  header <- if (distinct) "Distinct Matches" else "All Values"

  lines <- c(
    paste0("**** ", header, " ****"),
    "",
    "X in Y",
    sprintf("Of the %d X values, %d (%s%%) were matched.",
            total_x, xiny, format(pct_x, nsmall = 2)),
    strrep("*", 40),
    "",
    "Y in X",
    sprintf("Of the %d Y values, %d (%s%%) were matched.",
            total_y, yinx, format(pct_y, nsmall = 2)),
    strrep("*", 38)
  )

  writeLines(lines)
  invisible(NULL)
}