R/regexpr.R

In YuLab-SMU/yulab.utils: Supporting Functions for Packages Maintained by 'YuLab-SMU'

##' @rdname regexpr-style
##' @export
set_PCRE <- function() {
    options(regexpr_use_perl = TRUE)
}

##' @rdname regexpr-style
##' @export
set_TRE <- function() {
    options(regexpr_use_perl = FALSE)
}

##' @rdname regexpr-style
##' @export
use_perl <- function() {
    res <- getOption("regexpr_use_perl", default = auto_set_regexpr_style())
    return(res)
}


##' switch regular expression style (PCRE vs TRE)
##'
##' The `set_regexpr_style()` allows user to specify which style to be used, 
##' while the `auto_set_regexpr_style()` automatically set the style depdending on 
##' the operating system (TRE for Windows and PCRE for other OSs (Linux and Mac)).
##' 
##' `set_PCRE()` force to use PCRE style while `set_TRE()` force to use TRE. 
##' 
##' Note that all these functions are not change the behavior of `gsub()` and `regexpr()`. 
##' The functions are just set a global option to store the user's choice of whether using `perl = TRUE`.
##'
##' Users can access the option via `use_perl()` and pass the return value to `gusb()` or `regexpr()` to specify the style in use.
##' 
##' @rdname regexpr-style
##' @param style one of 'PCRE' or 'TRE'
##' @return logical value of whether use perl
##' @references <https://stackoverflow.com/questions/47240375/regular-expressions-in-base-r-perl-true-vs-the-default-pcre-vs-tre>
##' @export
##' @author Guangchuang Yu
set_regexpr_style <- function(style) {
    if (missing(style)) {
        message("style is not specific, set automatically.")
        auto_set_regexpr_style()
    } else {
        style <- match.arg(style, c("PCRE", "TRE"))
        if (style == "PCRE") {
            set_PCRE()
        } else {
            set_TRE()
        }
    }
    res <- getOption("regexpr_use_perl")
    invisible(res)
}

##' @rdname regexpr-style
##' @export
auto_set_regexpr_style <- function() {
    if (which_os() == "Windows") {
        set_TRE()
        res <- FALSE
    } else {
        set_PCRE()
        res <- TRUE
    }
    invisible(res)
}