this runs through nzpcn.org.nz and gets native range and naturalised year. used to run: w<-matrix(rep(1:7700)) test <- sapply(w, function(x) nat.year.nzpcn(x))

nat.year.nzpcn.R

nat.year.nzpcn <- function(x) {
    sp <- getURL(paste0("http://nzpcn.org.nz/flora_details.aspx?ID=", x))
    if (length(grep("Error: Cannot find species with ID =", sp))) {
        return("Not_found")
    } else {
        flush.console()
        sp.st <- unlist(gregexpr(pattern = "<title>", sp, ignore.case = F, 
                                 fixed = T))
        sp.end <- unlist(gregexpr(pattern = " | New Zealand Plant Conservation Network", 
                                  sp, ignore.case = F, fixed = T))
        species <- unique(substr(sp, sp.st + 9, sp.end))
        
        nat <- unlist(gregexpr(pattern = "Year Naturalised", sp, ignore.case = F, 
                               fixed = T))
        org <- unlist(gregexpr(pattern = "Origin", sp, ignore.case = F, 
                               fixed = T))
        
        if ((min(nat) <= 0) && (min(org) <= 0)) {
            results <- c(species, "NA", "NA")
        }
        if ((min(nat) >= 0) && (min(org) <= 0)) {
            nat.year <- unique(substr(sp, nat + 24, nat + 27))
            results <- c(species, nat.year, "NA")
        }
        if ((min(nat) <= 0) && (min(org) >= 0)) {
            org.loc <- unique(substr(sp, org + 14, org + 47))
            org.loc <- sub("</p>.*", "", org.loc)
            results <- c(species, "NA", org.loc)
        } else {
            nat.year <- unique(substr(sp, nat + 24, nat + 27))
            org.loc <- unique(substr(sp, org + 14, org + 47))
            org.loc <- sub("</p>.*", "", org.loc)
            results <- c(species, nat.year, org.loc)
        }
    }
}