remove unicode escape

remove_unicode_escape.R

source("hex2str.R")
remove_unicode_escape <- function(src) {
	start.set <- gregexpr("\\\\u[0-9a-fA-F]{4,4}", src, perl=FALSE, useBytes=TRUE)[[1]]
	sub.list <- list()
	for(start in start.set[length(start.set):1]) {
		if (!is.null(sub.list[[substring(src, start, start+5)]])) next
		pattern <- substring(src, first=start, last=start + 5)
		hex <- substring(pattern, 3, 6)
		result <- iconv(hex2str(hex), "UTF16BE", "UTF8")
		cat(paste(hex, "-->", result, "\n"))
		if (is.na(result)) stop(result)
		sub.list[[substring(src, start, start+5)]] <- result
	}
	for(i in 1:length(sub.list)) {
		src <- sub(pattern=names(sub.list)[i], replacement=sub.list[[i]], x=src, fixed=TRUE, useBytes=FALSE)
	}
	return(src)
}