dplyr-joins.R

#' Left join two data frames
#'
#' This is a `joyn` wrapper that works in a similar
#' fashion to [dplyr::left_join]
#'
#' @param x data frame: referred to as *left* in R terminology, or *master* in
#'   Stata terminology.
#' @param y data frame: referred to as *right* in R terminology, or *using* in
#'   Stata terminology.
#' @param by a character vector of variables to join by. If NULL, the default,
#'   joyn will do a natural join, using all variables with common names across
#'   the two tables. A message lists the variables so that you can check they're
#'   correct (to suppress the message, simply explicitly list the variables that
#'   you want to join). To join by different variables on x and y use a vector
#'   of expressions. For example, `by = c("a = b", "z")` will use "a" in `x`, "b"
#'   in `y`, and "z" in both tables.
#' @inheritParams dplyr::left_join
#' @inheritParams joyn
#' @inheritDotParams joyn
#'
#' @family dplyr alternatives
#' @return An data frame of the same class as `x`. The properties of the output
#' are as close as possible to the ones returned by the dplyr alternative.
#' @export
#'
#' @examples
#' # Simple left join
#' library(data.table)
#'
#' x1 = data.table(id = c(1L, 1L, 2L, 3L, NA_integer_),
#'                 t  = c(1L, 2L, 1L, 2L, NA_integer_),
#'                 x  = 11:15)
#' y1 = data.table(id = c(1,2, 4),
#'                 y  = c(11L, 15L, 16))
#' left_join(x1, y1, relationship = "many-to-one")
left_join <- function(
    x,
    y,
    by               = intersect(names(x), names(y)),
    copy             = FALSE,
    suffix           = c(".x", ".y"),
    keep             = NULL,
    na_matches       = c("na", "never"),
    multiple         = "all",
    unmatched        = "drop",
    relationship     = NULL,
    y_vars_to_keep   = TRUE,
    update_values    = FALSE,
    update_NAs       = update_values,
    reportvar        = getOption("joyn.reportvar"),
    reporttype       = c("character", "numeric"),
    roll             = NULL,
    keep_common_vars = FALSE,
    sort             = TRUE,
    verbose          = getOption("joyn.verbose"),
    ...
) {

  clear_joynenv()

  # Argument checks ---------------------------------
  x <- copy(x)
  y <- copy(y)
  na_matches <- match.arg(na_matches,
                          choices = c("na","never"))
  multiple   <- match.arg(multiple,
                          choices = c("all",
                                      "any",
                                      "first",
                                      "last"))
  unmatched  <- match.arg(unmatched,
                          choices = c("drop",
                                      "error"))

  args_check <- arguments_checks(x             = x,
                                 y             = y,
                                 by            = by,
                                 copy          = copy,
                                 keep          = keep,
                                 suffix        = suffix,
                                 na_matches    = na_matches,
                                 multiple      = multiple,
                                 relationship  = relationship,
                                 reportvar     = reportvar)
  by           <- args_check$by
  keep         <- args_check$keep
  na_matches   <- args_check$na_matches
  multiple     <- args_check$multiple
  relationship <- args_check$relationship
  reportvar    <- args_check$reportvar
  dropreport   <- args_check$dropreport

  # Column names -----------------------------------
  if (keep == TRUE) {
    jn_type <- "left"
    modified_cols <- set_col_names(x       = x,
                                   y       = y,
                                   by      = by,
                                   jn_type = jn_type,
                                   suffix  = suffix)
    x <- modified_cols$x
    y <- modified_cols$y
  }

  # Execute left join------------------------------------
  lj <- joyn(
    x                = x,
    y                = y,
    by               = by,
    match_type       = relationship,
    keep             = "left",
    y_vars_to_keep   = y_vars_to_keep,
    suffixes         = suffix,
    update_values    = update_values,
    update_NAs       = update_NAs,
    reportvar        = reportvar,
    reporttype       = reporttype,
    keep_common_vars = TRUE,
    sort             = sort,
    verbose          = verbose,
    ...
  )

  # Unmatched Keys ---------------------------------------
  if (unmatched == "error") {
    check_unmatched_keys(x       = x,
                         y       = y,
                         out     = lj,
                         by      = by,
                         jn_type = "left")
  }
  # Should report be kept---------------------------------
  if (dropreport == T) {
    get_vars(lj, reportvar) <- NULL
  }

  # return
  lj
}

#' Right join two data frames
#'
#' This is a `joyn` wrapper that works in a similar
#' fashion to [dplyr::right_join]
#'
#' @param x data frame: referred to as *left* in R terminology, or *master* in
#'   Stata terminology.
#' @param y data frame: referred to as *right* in R terminology, or *using* in
#'   Stata terminology.
#' @param by a character vector of variables to join by. If NULL, the default,
#'   joyn will do a natural join, using all variables with common names across
#'   the two tables. A message lists the variables so that you can check they're
#'   correct (to suppress the message, simply explicitly list the variables that
#'   you want to join). To join by different variables on x and y use a vector
#'   of expressions. For example, `by = c("a = b", "z")` will use "a" in `x`, "b"
#'   in `y`, and "z" in both tables.
#' @inheritParams dplyr::right_join
#' @inheritParams joyn
#' @inheritDotParams joyn
#'
#' @family dplyr alternatives
#' @inherit left_join return
#' @export
#'
#' @examples
#' # Simple right join
#' library(data.table)
#'
#' x1 = data.table(id = c(1L, 1L, 2L, 3L, NA_integer_),
#'                 t  = c(1L, 2L, 1L, 2L, NA_integer_),
#'                 x  = 11:15)
#' y1 = data.table(id = c(1,2, 4),
#'                 y  = c(11L, 15L, 16))
#' right_join(x1, y1, relationship = "many-to-one")
right_join <- function(
    x,
    y,
    by               = intersect(names(x), names(y)),
    copy             = FALSE,
    suffix           = c(".x", ".y"),
    keep             = NULL,
    na_matches       = c("na", "never"),
    multiple         = "all",
    unmatched        = "drop",
    relationship     = "one-to-one",
    y_vars_to_keep   = TRUE,
    update_values    = FALSE,
    update_NAs       = update_values,
    reportvar        = getOption("joyn.reportvar"),
    reporttype       = c("character", "numeric"),
    roll             = NULL,
    keep_common_vars = FALSE,
    sort             = TRUE,
    verbose          = getOption("joyn.verbose"),
    ...
) {

  clear_joynenv()

  # Argument checks ---------------------------------
  x <- copy(x)
  y <- copy(y)
  na_matches <- match.arg(na_matches,
                          choices = c("na","never"))
  multiple   <- match.arg(multiple,
                          choices = c("all",
                                      "any",
                                      "first",
                                      "last"))
  unmatched  <- match.arg(unmatched,
                          choices = c("drop",
                                      "error"))

  args_check <- arguments_checks(x             = x,
                                 y             = y,
                                 by            = by,
                                 copy          = copy,
                                 keep          = keep,
                                 suffix        = suffix,
                                 na_matches    = na_matches,
                                 multiple      = multiple,
                                 relationship  = relationship,
                                 reportvar     = reportvar)
  by           <- args_check$by
  keep         <- args_check$keep
  na_matches   <- args_check$na_matches
  multiple     <- args_check$multiple
  relationship <- args_check$relationship
  reportvar    <- args_check$reportvar
  dropreport   <- args_check$dropreport

  # Column names -----------------------------------
  if (keep == TRUE) {
    jn_type <- "right"
    modified_cols <- set_col_names(x       = x,
                                   y       = y,
                                   by      = by,
                                   jn_type = jn_type,
                                   suffix  = suffix)
    x <- modified_cols$x
    y <- modified_cols$y
  }

  # Execute right join ------------------------------------
  rj <- joyn(
    x                = x,
    y                = y,
    by               = by,
    match_type       = relationship,
    keep             = "right",
    y_vars_to_keep   = y_vars_to_keep,
    suffixes         = suffix,
    update_values    = update_values,
    update_NAs       = update_NAs,
    reportvar        = reportvar,
    reporttype       = reporttype,
    keep_common_vars = TRUE,
    sort             = sort,
    verbose          = verbose,
    ...
  )

  # Unmatched Keys ---------------------------------------
  if (unmatched == "error") {
    check_unmatched_keys(x       = x,
                         y       = y,
                         out     = rj,
                         by      = by,
                         jn_type = "right")
  }

  # Should reportvar be kept
  if (dropreport == T) {
    get_vars(rj, reportvar) <- NULL
  }

  # Return
  rj

}

#' Full join two data frames
#'
#' This is a `joyn` wrapper that works in a similar
#' fashion to [dplyr::full_join]
#'
#' @param x data frame: referred to as *left* in R terminology, or *master* in
#'   Stata terminology.
#' @param y data frame: referred to as *right* in R terminology, or *using* in
#'   Stata terminology.
#' @param by a character vector of variables to join by. If NULL, the default,
#'   joyn will do a natural join, using all variables with common names across
#'   the two tables. A message lists the variables so that you can check they're
#'   correct (to suppress the message, simply explicitly list the variables that
#'   you want to join). To join by different variables on x and y use a vector
#'   of expressions. For example, `by = c("a = b", "z")` will use "a" in `x`, "b"
#'   in `y`, and "z" in both tables.
#' @inheritParams dplyr::full_join
#' @inheritParams joyn
#' @inheritDotParams joyn
#'
#' @family dplyr alternatives
#'
#' @inherit left_join return
#' @export
#'
#' @examples
#' # Simple full join
#' library(data.table)
#'
#' x1 = data.table(id = c(1L, 1L, 2L, 3L, NA_integer_),
#'                 t  = c(1L, 2L, 1L, 2L, NA_integer_),
#'                 x  = 11:15)
#' y1 = data.table(id = c(1,2, 4),
#'                 y  = c(11L, 15L, 16))
#' full_join(x1, y1, relationship = "many-to-one")
full_join <- function(
    x,
    y,
    by               = intersect(names(x), names(y)),
    copy             = FALSE,
    suffix           = c(".x", ".y"),
    keep             = NULL,
    na_matches       = c("na", "never"),
    multiple         = "all",
    unmatched        = "drop",
    relationship     = "one-to-one",
    y_vars_to_keep   = TRUE,
    update_values    = FALSE,
    update_NAs       = update_values,
    reportvar        = getOption("joyn.reportvar"),
    reporttype       = c("character", "numeric"),
    roll             = NULL,
    keep_common_vars = FALSE,
    sort             = TRUE,
    verbose          = getOption("joyn.verbose"),
    ...
) {

  clear_joynenv()

  # Argument checks ---------------------------------
  x <- copy(x)
  y <- copy(y)
  na_matches <- match.arg(na_matches,
                          choices = c("na","never"))
  multiple   <- match.arg(multiple,
                          choices = c("all",
                                      "any",
                                      "first",
                                      "last"))
  unmatched  <- match.arg(unmatched,
                          choices = c("drop",
                                      "error"))

  args_check <- arguments_checks(x             = x,
                                 y             = y,
                                 by            = by,
                                 copy          = copy,
                                 keep          = keep,
                                 suffix        = suffix,
                                 na_matches    = na_matches,
                                 multiple      = multiple,
                                 relationship  = relationship,
                                 reportvar     = reportvar)
  by           <- args_check$by
  keep         <- args_check$keep
  na_matches   <- args_check$na_matches
  multiple     <- args_check$multiple
  relationship <- args_check$relationship
  reportvar    <- args_check$reportvar
  dropreport   <- args_check$dropreport

  # Column names -----------------------------------
  if (keep == TRUE) {
    jn_type <- "full"
    modified_cols <- set_col_names(x       = x,
                                   y       = y,
                                   by      = by,
                                   jn_type = jn_type,
                                   suffix  = suffix)
    x <- modified_cols$x
    y <- modified_cols$y
  }


  # Execute full join ------------------------------------
  fj <- joyn(
    x                = x,
    y                = y,
    by               = by,
    match_type       = relationship,
    keep             = "full",
    y_vars_to_keep   = y_vars_to_keep,
    suffixes         = suffix,
    update_values    = update_values,
    update_NAs       = update_NAs,
    reportvar        = reportvar,
    reporttype       = reporttype,
    keep_common_vars = TRUE,
    sort             = sort,
    verbose          = verbose,
    ...
  )

  # Unmatched Keys----------------------------------------
  if (unmatched == "error") {

    # Store warning message
    store_msg(
      type        = "warn",
      warn        = paste(cli::symbol$warn, "\nWarning:"),
      pale        = "\nargument",
      bolded_pale = "  warning = error",
      pale        = "\nis not active in this type of",
      bolded_pale = "  joyn"
    )

  }

  # Should reportvar be kept
  if (dropreport == T) {
    get_vars(fj, reportvar) <- NULL
  }

  # Return
  fj

}


#' Inner join two data frames
#'
#' This is a `joyn` wrapper that works in a similar fashion to
#' [dplyr::inner_join]
#'
#' @param x data frame: referred to as *left* in R terminology, or *master* in
#'   Stata terminology.
#' @param y data frame: referred to as *right* in R terminology, or *using* in
#'   Stata terminology.
#' @param by a character vector of variables to join by. If NULL, the default,
#'   joyn will do a natural join, using all variables with common names across
#'   the two tables. A message lists the variables so that you can check they're
#'   correct (to suppress the message, simply explicitly list the variables that
#'   you want to join). To join by different variables on x and y use a vector
#'   of expressions. For example, `by = c("a = b", "z")` will use "a" in `x`,
#'   "b" in `y`, and "z" in both tables.
#' @inheritParams dplyr::inner_join
#' @inheritParams joyn
#' @inheritDotParams joyn
#'
#' @family dplyr alternatives
#' @export
#' @inherit left_join return
#' @examples
#' # Simple full join
#' library(data.table)
#'
#' x1 = data.table(id = c(1L, 1L, 2L, 3L, NA_integer_),
#'                 t  = c(1L, 2L, 1L, 2L, NA_integer_),
#'                 x  = 11:15)
#' y1 = data.table(id = c(1,2, 4),
#'                 y  = c(11L, 15L, 16))
#' inner_join(x1, y1, relationship = "many-to-one")
inner_join <- function(
    x,
    y,
    by               = intersect(names(x), names(y)),
    copy             = FALSE,
    suffix           = c(".x", ".y"),
    keep             = NULL,
    na_matches       = c("na", "never"),
    multiple         = "all",
    unmatched        = "drop",
    relationship     = "one-to-one",
    y_vars_to_keep   = TRUE,
    update_values    = FALSE,
    update_NAs       = update_values,
    reportvar        = getOption("joyn.reportvar"),
    reporttype       = c("character", "numeric"),
    roll             = NULL,
    keep_common_vars = FALSE,
    sort             = TRUE,
    verbose          = getOption("joyn.verbose"),
    ...
) {

  clear_joynenv()

  # Argument checks ---------------------------------
  x <- copy(x)
  y <- copy(y)
  na_matches <- match.arg(na_matches,
                          choices = c("na","never"))
  multiple   <- match.arg(multiple,
                          choices = c("all",
                                      "any",
                                      "first",
                                      "last"))
  unmatched  <- match.arg(unmatched,
                          choices = c("drop",
                                      "error"))

  args_check <- arguments_checks(x             = x,
                                 y             = y,
                                 by            = by,
                                 copy          = copy,
                                 keep          = keep,
                                 suffix        = suffix,
                                 na_matches    = na_matches,
                                 multiple      = multiple,
                                 relationship  = relationship,
                                 reportvar     = reportvar)
  by           <- args_check$by
  keep         <- args_check$keep
  na_matches   <- args_check$na_matches
  multiple     <- args_check$multiple
  relationship <- args_check$relationship
  reportvar    <- args_check$reportvar
  dropreport   <- args_check$dropreport

  # Column names -----------------------------------
  if (keep == TRUE) {
    jn_type <- "inner"
    modified_cols <- set_col_names(x       = x,
                                   y       = y,
                                   by      = by,
                                   jn_type = jn_type,
                                   suffix  = suffix)
    x <- modified_cols$x
    y <- modified_cols$y
  }

  # Execute inner join ------------------------------------
  ij <- joyn(
    x                = x,
    y                = y,
    by               = by,
    match_type       = relationship,
    keep             = "inner",
    y_vars_to_keep   = y_vars_to_keep,
    suffixes         = suffix,
    update_values    = update_values,
    update_NAs       = update_NAs,
    reportvar        = reportvar,
    reporttype       = reporttype,
    keep_common_vars = TRUE,
    sort             = sort,
    verbose          = verbose,
    ...
  )

  # Unmatched Keys ---------------------------------------
  if (unmatched == "error") {
    check_unmatched_keys(x       = x,
                         y       = y,
                         out     = ij,
                         by      = by,
                         jn_type = "inner")
  }

  ### if dropreport = T
  if (dropreport == T) {
    get_vars(ij, reportvar) <- NULL
  }

  # Return
  ij

}


# HELPER FUNCTIONS -------------------------------------------------------------
## Arguments checks ####

#' Perform necessary preliminary checks on arguments that are passed to joyn
#' @param x data frame: left table
#' @param y data frame: right table
#' @param by character vector or variables to join by
#' @inheritParams left_join
#' @return list of checked arguments to pass on to the main joyn function
#' @keywords internal
arguments_checks <- function(x, y, by, copy, keep, suffix, na_matches, multiple,
                             relationship, reportvar) {
  # Check by
  if (is.null(by)) {
    by <- intersect(
      names(x),
      names(y)
    )
  }

  # Check copy
  if (copy == TRUE) {
    store_msg(
      type        = "warn",
      warn        = paste(cli::symbol$warn, "\nWarning:"),
      pale        = "\nargument",
      bolded_pale = "  copy = TRUE",
      pale        = "\nis not active in this version of",
      bolded_pale = "  joyn"
    )
  }

  # Check suffix
  if (is.null(suffix) || !length(suffix) == 2 || !is.character(suffix)) {
    cli::cli_abort(
      paste0(
        cli::symbol$cross,
        " Error: argument `suffix` must be character vector of length 2"
      )
    )
  }

  # Check keep
  if (!is.null(keep) & !is.logical(keep)) {
    cli::cli_abort(
      paste0(
        cli::symbol$cross,
        " Error: argument `keep` should be one of NULL, TRUE, or FALSE"
      )
    )
  }
  if (is.null(keep)) {
    store_msg(
      type        = "warn",
      warn        = paste(cli::symbol$warn,"\nWarning:"),
      pale        = "  joyn does not currently allow inequality joins, so",
      bolded_pale = "  keep = NULL",
      pale        = "  will retain only keys in x"
    )
    keep <- FALSE
  }

  # Check multiple
  if (multiple == "any") {
    multiple <- "first"
  }

  # Check relationship
  if (is.null(relationship)) {relationship <- "one-to-one"}

  relationship <- switch(
    relationship,
    "one-to-one"   = "1:1",
    "one-to-many"  = "1:m",
    "many-to-one"  = "m:1",
    "many-to-many" = "m:m"
  )
  if (
    relationship %in% c("1:m", "m:m") &
    !multiple == "all"
  ) {
    cli::cli_abort(
      paste0(
        cli::symbol$cross,
        " Error: if `relationship` is 1:m or m:m then `multiple` should be 'all' "
      )
    )
  }

  # Check na_matches
  if (na_matches == "never") {
    store_msg(
      type        = "warn",
      warn        = paste(cli::symbol$warn, "\nWarning:"),
      pale        = "  Currently, joyn allows only",
      bolded_pale = "  na_matches = 'na'"
    )
  }

  # Check reportvar
  if (is.null(reportvar) || isFALSE(reportvar)) {
    dropreport <- TRUE
    reportvar <- getOption("joyn.reportvar")
  } else{
    dropreport <- FALSE
  }

  out <- list(by           = by,
              copy         = copy,
              suffix       = suffix,
              keep         = keep,
              na_matches   = na_matches,
              multiple     = multiple,
              relationship = relationship,
              reportvar    = reportvar,
              dropreport   = dropreport)

  return(out)

}


#' Add x key var and y key var (with suffixes) to x and y
#' -when joining by different variables and keep is true
#' @param x data table: left table
#' @param y data table: right table
#' @param by character vector of variables to join by
#' @param suffix character(2) specifying the suffixes to be used for making non-by column names unique
#' @param jn_type character specifying type of join
#' @return list containing x and y
#' @keywords internal
set_col_names <- function(x, y, by, suffix, jn_type) {

  x_1 <- copy(x)
  y_1 <- copy(y)

  # If joining by different variables
  if (length(grep(pattern = "==?", x = by, value = TRUE)) != 0) {

    if (jn_type == "right") {
      by_x_names <- fix_by_vars(by = by, x_1, y_1)$xby
    }

    else if (jn_type == "left" | jn_type == "full" | jn_type == "inner") {
      by_y_names <- fix_by_vars(by = by, x_1, y_1)$yby
    }

  }

  # If joining by common var
  else {
    by_y_names <- by_x_names <- fix_by_vars(by = by, x_1, y_1)$by
    }

  # Add key vars with suffix to x and y
  if (jn_type == "right") {
    xkeys <- x |>
      fselect(by_x_names)
    names(xkeys) <- paste0(names(xkeys), suffix[1])
    x <- cbind(
      xkeys,
      x
    )
  } else if (jn_type == "left" | jn_type == "full" | jn_type == "inner")  {

    ykeys <- y |>
      fselect(by_y_names)
    names(ykeys) <- paste0(names(ykeys), suffix[2])
    y <- cbind(
      ykeys,
      y
    )

  } #close else

  return(list(x = x,
              y = y))

}



#' Conduct all unmatched keys checks and return error if necessary
#'
#' @param x left table
#' @param y right table
#' @param out output from join
#' @param by character vector of keys that x and y are joined by
#' @param jn_type character: "left", "right", or "inner"
#'
#' @return error message
#' @keywords internal
check_unmatched_keys <- function(x, y, out, by, jn_type) {

  # Left table --------------------------------------------------------
  if (jn_type == "left" | jn_type == "inner") {

    use_y_input <- process_by_vector(by = by, input = "right")
    use_y_out   <- process_by_vector(by = by, input = "left")

      if (length(grep("==?", by, value = TRUE)) != 0) {

        if (any(use_y_out %in% colnames(y))) {

          store_msg(
            type         = "warn",
            warn         = paste(cli::symbol$warn, "\nWarning:"),
            pale         = "\nUnmatched = error not active for this joyn -unmatched keys are not detected"
          )
        }

        else {
          data.table::setnames(y, new = use_y_out, old = use_y_input)

          if (unmatched_keys(x         = y,
                                 by        = use_y_out,
                                 out       = out)) {
            cli::cli_abort(
              paste0(
                cli::symbol$cross,
                " Error: some rows in `y` are not matched - this check is due to
           argument `unmatched = 'error'` ")
            )
          }
        }
      }

      else {
        if (unmatched_keys(x         = y,
                               by        = use_y_out,
                               out       = out)) {
          cli::cli_abort(
            paste0(
              cli::symbol$cross,
              " Error: some rows in `y` are not matched - this check is due to
           argument `unmatched = 'error'` ")
          )
        }
      }

    }


  # Right Join --------------------------------------------------------
  if (jn_type == "right" | jn_type == "inner") {

    use_x_input <- process_by_vector(by = by, input = "left")

      if (unmatched_keys(x         = x,
                             by        = use_x_input,
                             out       = out)) {
        cli::cli_abort(
          paste0(
            cli::symbol$cross,
            " Error: some rows in `x` are not matched - this check is due to
           argument `unmatched = 'error'`. To drop these rows, set `unmatched = 'drop'` ")
        )
      }
  }

  invisible(x)

}


#' Check for unmatched keys
#'
#' Gives TRUE if unmatched keys, FALSE if not.
#'
#' @param x input table to join
#' @param out output of join
#' @param by by argument, giving keys for join
#'
#' @return logical
#' @keywords internal
unmatched_keys <- function(x, out, by) {

  check <- NULL

  # Get all keys from `x`
  x_keys <- x |>
    fselect(by) |>
    as.data.table()

  # get all key combos from `out`
  out_keys <- out |>
    fselect(by) |>
    as.data.table()

  # check that key combos are equal
  check <- (data.table::fsetdiff(x_keys,
                                 out_keys) |>
              nrow()) > 0  # if true  => more unique combos in x
  #    false => same unique combos of keys
  # same number unique keys =>
  #     all matched keys
  #     because output is result
  #     of join
  check
}



#' Process the `by` vector
#'
#' Gives as output a vector of names to be used for the specified
#' table that correspond to the `by` argument for that table
#'
#' @param by character vector: by argument for join
#' @param input character: either "left" or "right", indicating
#' whether to give the left or right side of the equals ("=") if
#' the equals is part of the `by` vector
#'
#' @return character vector
#' @keywords internal
#'
#' @examples
#' joyn:::process_by_vector(by = c("An = foo", "example"), input = "left")
process_by_vector <- function(by, input = c("left", "right")) {
  input <- match.arg(input)
  if (input == "left") {
    out <- sapply(by, function(x) {
      if (grepl("=", x)) {
        trimws(gsub("([^=]+)(\\s*==?\\s*)([^=]+)", "\\1", x))
      } else {
        x
      }
    })
  } else if (input == "right") {
    out <- sapply(by, function(x) {
      if (grepl("=", x)) {
        trimws(gsub("([^=]+)(\\s*==?\\s*)([^=]+)", "\\3", x))
      } else {
        x
      }
    })
  }
  out |> unname()
}