Download zipped HUC 6-digit HAND packages from CFIM Data Repository

download_hand_cfim.R

#' Download from CFIM
#' Download zipped HUC 6-digit HAND packages from CFIM Data Repository - https://cfim.ornl.gov/data/
#' @param huc6_codes character vector of HUC 6-digit codes, or "all" to download all available HUC 6-digit zip files.
#' @param dir character of directory path.
#' @param overwrite Overwrite already downloaded zip files? Defaults to FALSE.
#' @return No explicit return. Unzipped HUC 6-digit HAND packages can be found in `dir`
#' importsFrom RCurl getURL
#' importsFrom stringr str_extract_all
#' importsFrom magrittr %>%
download_hand_cfim <- function(huc6_codes = "all", dir, overwrite = FALSE, ...) {
    # Create directory if it doesn't exist
    if (!dir.exists(dir)) dir.create(dir)

    # Get vector of HUC6 codes
    all_huc6_codes <- scan("https://cfim.ornl.gov/data/list.HUC6.txt",
                      character(),
                      quote = "")
    
    # Get all downloadable HUC6 codes via RCurl and subset
    all_huc6_codes <- RCurl::getURL("https://cfim.ornl.gov/data/HAND/20200601/") %>%
                      stringr::str_extract_all("<a [^>]+>(.*?)</a>") %>%
                      sapply(substr, start = 22, stop = 27)
    
    # Cast to vector and remove misc data
    all_huc6_codes <- all_huc6_codes[, 1][-c(1:5)]
    
    if (huc6_codes == "all") huc6_codes <- all_huc6_codes
    
    # Don't redo ones already downloaded
    if (!overwrite) {
        huc6_codes <- huc6_codes[
            !(huc6_codes %in% unname(sapply(
                list.files("data/pull/"),
                substr,
                start = 0,
                stop = 6
            )))
        ]
    }

    # All HUC6 codes in "huc6_codes"
    for (huc6_code in huc6_codes) {

        if (!(huc6_code %in% all_huc6_codes)) {
            warning(huc6_code, " not downloadable. Ensure it is valid. Skipping...")
            next
        }

        # Create temp file
        temp  <- tempfile(
            pattern = paste0(huc6_code, "-"),
            fileext = ".zip",
            tmpdir = dir
        )

        tryCatch({
            # Download package from CFIM Data Repository
            download.file(
                paste0(
                    "https://cfim.ornl.gov/data/HAND/20200601/",
                    huc6_code,
                    ".zip"
                ),
                temp,
                method = "libcurl"
            )

            unzip(temp, exdir = dir)

            unlink(temp)
        },
        error = function(cond) {
            message("Error at HUC6: ", huc6_code)
            stop(cond)
        })
    }
}