Nothing
# Wrapping vctrs data_frame constructor with no name repair
data_frame0 <- function(...) data_frame(..., .name_repair = "minimal")
# Wrapping unique0() to accept NULL
unique0 <- function(x, ...) if (is.null(x)) x else vec_unique(x, ...)
df_rows <- function(x, i) {
cols <- lapply(x, `[`, i = i)
data_frame0(!!!cols, .size = length(i))
}
split_matrix <- function(x, col_names = colnames(x)) {
force(col_names)
x <- lapply(seq_len(ncol(x)), function(i) x[, i])
if (!is.null(col_names)) names(x) <- col_names
x
}
# More performant modifyList without recursion
modify_list <- function(old, new) {
for (i in names(new)) old[[i]] <- new[[i]]
old
}
empty <- function(df) {
is.null(df) || nrow(df) == 0 || ncol(df) == 0
}
split_indices <- function(group) {
split(seq_along(group), group)
}
# Adapted from plyr:::id_vars
# Create a unique id for elements in a single vector
id_var <- function(x, drop = FALSE) {
if (length(x) == 0) {
id <- integer()
n <- 0L
} else if (!is.null(attr(x, 'n')) && !drop) {
return(x)
} else if (is.factor(x) && !drop) {
x <- addNA(x, ifany = TRUE)
id <- as.integer(x)
n <- length(levels(x))
} else {
levels <- sort(unique0(x), na.last = TRUE)
id <- match(x, levels)
n <- max(id)
}
attr(id, 'n') <- n
id
}
#' Create an unique integer id for each unique row in a data.frame
#'
#' Properties:
#' - `order(id)` is equivalent to `do.call(order, df)`
#' - rows containing the same data have the same value
#' - if `drop = FALSE` then room for all possibilites
#'
#' @param .variables list of variables
#' @param drop Should unused factor levels be dropped?
#'
#' @return An integer vector with attribute `n` giving the total number of
#' possible unique rows
#'
#' @keywords internal
#' @noRd
#'
id <- function(.variables, drop = FALSE) {
nrows <- NULL
if (is.data.frame(.variables)) {
nrows <- nrow(.variables)
.variables <- unclass(.variables)
}
lengths <- vapply(.variables, length, integer(1))
.variables <- .variables[lengths != 0]
if (length(.variables) == 0) {
n <- nrows %||% 0L
id <- seq_len(n)
attr(id, 'n') <- n
return(id)
}
if (length(.variables) == 1) {
return(id_var(.variables[[1]], drop = drop))
}
ids <- rev(lapply(.variables, id_var, drop = drop))
p <- length(ids)
ndistinct <- vapply(ids, attr, 'n', FUN.VALUE = numeric(1), USE.NAMES = FALSE)
n <- prod(ndistinct)
if (n > 2^31) {
char_id <- inject(paste(!!!ids, sep = '\r'))
res <- match(char_id, unique0(char_id))
}
else {
combs <- c(1, cumprod(ndistinct[-p]))
mat <- inject(cbind(!!!ids))
res <- c((mat - 1L) %*% combs + 1L)
}
if (drop) {
id_var(res, drop = TRUE)
}
else {
res <- as.integer(res)
attr(res, 'n') <- n
res
}
}
#' Apply function to unique subsets of a data.frame
#'
#' This function is akin to `plyr::ddply`. It takes a single data.frame,
#' splits it by the unique combinations of the columns given in `by`, apply a
#' function to each split, and then reassembles the results into a sigle
#' data.frame again.
#'
#' @param df A data.frame
#' @param by A character vector of column names to split by
#' @param fun A function to apply to each split
#' @param ... Further arguments to `fun`
#' @param drop Should unused factor levels in the columns given in `by` be
#' dropped.
#'
#' @return A data.frame if the result of `fun` does not include the columns
#' given in `by` these will be prepended to the result.
#'
#' @keywords internal
#' @importFrom stats setNames
#' @noRd
dapply <- function(df, by, fun, ..., drop = TRUE) {
grouping_cols <- .subset(df, by)
fallback_order <- unique0(c(by, names(df)))
apply_fun <- function(x) {
res <- fun(x, ...)
if (is.null(res)) return(res)
if (length(res) == 0) return(data_frame0())
vars <- lapply(setNames(by, by), function(col) .subset2(x, col)[1])
if (is.matrix(res)) res <- split_matrix(res)
if (is.null(names(res))) names(res) <- paste0("V", seq_along(res))
if (all(by %in% names(res))) return(data_frame0(!!!unclass(res)))
res <- modify_list(unclass(vars), unclass(res))
res <- res[intersect(c(fallback_order, names(res)), names(res))]
data_frame0(!!!res)
}
# Shortcut when only one group
if (all(vapply(grouping_cols, single_val, logical(1)))) {
return(apply_fun(df))
}
ids <- id(grouping_cols, drop = drop)
group_rows <- split_with_index(seq_len(nrow(df)), ids)
result <- lapply(seq_along(group_rows), function(i) {
cur_data <- df_rows(df, group_rows[[i]])
apply_fun(cur_data)
})
vec_rbind(!!!result)
}
# Use chartr() for safety since toupper() fails to convert i to I in Turkish locale
lower_ascii <- "abcdefghijklmnopqrstuvwxyz"
upper_ascii <- "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
to_lower_ascii <- function(x) chartr(upper_ascii, lower_ascii, x)
to_upper_ascii <- function(x) chartr(lower_ascii, upper_ascii, x)
tolower <- function(x) {
cli::cli_abort("Please use {.fn to_lower_ascii}, which works fine in all locales.")
}
toupper <- function(x) {
cli::cli_abort("Please use {.fn to_upper_ascii}, which works fine in all locales.")
}
# Convert a snake_case string to camelCase
camelize <- function(x, first = FALSE) {
x <- gsub("_(.)", "\\U\\1", x, perl = TRUE)
if (first) x <- firstUpper(x)
x
}
snakeize <- function(x) {
x <- gsub("([A-Za-z])([A-Z])([a-z])", "\\1_\\2\\3", x)
x <- gsub(".", "_", x, fixed = TRUE)
x <- gsub("([a-z])([A-Z])", "\\1_\\2", x)
to_lower_ascii(x)
}
firstUpper <- function(s) {
paste0(to_upper_ascii(substring(s, 1, 1)), substring(s, 2))
}
snake_class <- function(x) {
snakeize(class(x)[1])
}
single_val <- function(x, ...) {
UseMethod("single_val")
}
#' @export
single_val.default <- function(x, ...) {
# This is set by id() used in creating the grouping var
identical(attr(x, "n"), 1L)
}
#' @export
single_val.factor <- function(x, ...) {
# Panels are encoded as factor numbers and can never be missing (NA)
identical(levels(x), "1")
}
with_seed_null <- function(seed, code) {
if (is.null(seed)) {
code
} else {
withr::with_seed(seed, code)
}
}
Any scripts or data that you put into this service are public.
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.