R/urls_filter.R

In kongdd/CMIP5tools: Functions to Tidy and Analysis CMIP5/6 Data

#' @importFrom lubridate make_date
#' @export
urls_filter <- function(files, mip_era = "CMIP6", period = NULL, ...){
    DATE_MAX <- "2101-01-01" # for rcp
    scenario <- get_scenario(files[1])
    is_rcp <- grepl("rcp|RCP|ssp|SSP", scenario)
    
    # filter ensembles
    inds <- seq_along(files)
    ensembles <- extract_ensemble(files)
    
    if (mip_era == "CMIP5") {
        if (!grepl("historicalMisc|misc|Misc", scenario)) {
            nchar = 6
            ind   <- which(substr(ensembles, 1, nchar) == substr("r1i1p1f1", 1, nchar))
            inds  <- inds[ind]
            files <- files[ind]
        }
    } else if (mip_era == "CMIP6") {
        nchar = 6 # r1i1p1f1, or r1i1p1
        ind   <- which(substr(ensembles, 1, nchar) == substr("r1i1p1f1", 1, nchar))
        inds  <- inds[ind]
        files <- files[ind]
    }

    if (is_rcp) period <- c(2006, 2100)
    dfile <- CMIP5Files_filter(files, duration = NULL, period = period, verbose = 1)
    # if (is_rcp) dfile <- dfile[ensemble == "r1i1p1", ]
    # if (is_rcp || grepl(scenario, "piControl|picontrol")) 
    
    if (is_rcp) {
        # this part is not used
        dfile_full <- CMIP5Files_filter(files, duration = NULL, verbose = 0)
        d_22   <- dfile_full[start >= DATE_MAX, ]
        cat(sprintf('[m] %s files out of 21th century ... \n', num_bad(nrow(d_22))))
        if (nrow(d_22) > 0) {
            ds <- CMIP5Files_summary(d_22)
            print(head(ds))
        }
    }

    ok(sprintf('[ok] %s duplicated file removed.\n', length(files) - nrow(dfile)))
    dfile
}