icd (version 3.2.0)

factor_split_na: (re-)factor and split into matched and unmatched elements

Description

I think this is now obsolete, now I have better integer-based re-factoring in C++.

Usage

factor_split_na(x, levels, factor_fun = factor_nosort_rcpp)

Details

Critically, this function does not convert factor to character vector and back to factor in order to modify factor levels, resulting in huge speed improvements for long vectors.

Examples

# NOT RUN {
icd:::factor_nosort_rcpp(c("1", NA)) # NA becomes a level
icd:::factor_nosort_rcpp(c("1", "2"), "1") # NA not a level, just dropped!
icd:::factor_nosort_rcpp(c("1", "2"), c("1", NA)) # NA IS a level
# }
# NOT RUN {
x <- c("A", "B", "C", "d", "A", "C")
levels <- c("A", "B")
stopifnot(
  identical(icd:::factor_split_na(factor(x), levels),
            icd:::factor_split_na(x, levels))
)
y <- c("A", NA, "B", "A", NA)
yf <- factor(y)
yf_na <- factor(y, levels = c("A", NA, "B"), exclude = NULL)
stopifnot(
  identical(icd:::factor_split_na(y, "A"),
            icd:::factor_split_na(yf, "A"))
)
stopifnot(
  identical(icd:::factor_split_na(y, "A"),
            icd:::factor_split_na(yf_na, "A"))
)
# }