R/pruneAppBinaries.R

In acidgenomics/r-koopa: Shell Bootloader for Bioinformatics

#' Prune app binary tarballs on AWS S3
#'
#' @export
#' @note Updated 2023-01-31.
#'
#' @param dryRun `logical(1)`.
#' Perform a dry run. Just return the S3 URLs, but don't delete anything.
#'
#' @examples
#' out <- pruneAppBinaries(dryRun = TRUE)
#' print(out)
pruneAppBinaries <- function(dryRun = FALSE) {
    assert(
        requireNamespaces("jsonlite"),
        isFlag(dryRun)
    )
    dict <- c(
        "bucket" = "private.koopa.acidgenomics.com",
        "profile" = "acidgenomics",
        "subdir" = "binaries"
    )
    alert(sprintf(
        "Pruning binaries in {.url %s}.",
        paste0("s3://", dict[["bucket"]], "/", dict[["subdir"]], "/")
    ))
    x <- shell(
        command = "aws",
        args = c(
            "--profile", dict[["profile"]],
            "s3api", "list-objects",
            "--bucket", dict[["bucket"]],
            "--output", "json"
        ),
        print = FALSE
    )
    x <- x[["stdout"]]
    json <- jsonlite::fromJSON(x)
    assert(
        is.list(json),
        isSubset("Contents", names(json))
    )
    df <- json[["Contents"]][, c("Key", "LastModified")]
    assert(is.data.frame(df))
    colnames(df) <- c("key", "lastModified")
    keep <- grepl(pattern = paste0("^", dict[["subdir"]], "/"), x = df[["key"]])
    df <- df[keep, , drop = FALSE]
    dupes <- unique(dirname(df[["key"]])[duplicated(dirname(df[["key"]]))])
    if (!hasLength(dupes)) {
        alertWarning("No binaries to prune.")
        return(invisible(character()))
    }
    keep <- dirname(df[, "key"]) %in% dupes
    df <- df[keep, ]
    ts <- df[["lastModified"]]
    ts <- sub(pattern = "T", replacement = " ", x = ts, fixed = TRUE)
    ts <- as.POSIXlt(ts)
    df[["lastModified"]] <- ts
    df[["dirname"]] <- dirname(df[["key"]])
    df <- df[order(df[["lastModified"]], decreasing = TRUE), ]
    spl <- split(x = df, f = df[["dirname"]])
    pruneKeys <- lapply(
        X = spl,
        FUN = function(x) {
            x[["key"]][seq(from = 2L, to = nrow(x))]
        }
    )
    pruneKeys <- unlist(pruneKeys, recursive = FALSE, use.names = FALSE)
    prunePaths <- paste0("s3://", dict[["bucket"]], "/", pruneKeys)
    if (isTRUE(dryRun)) {
        alertInfo("Dry run mode.")
        cat(prunePaths, sep = "\n")
        return(invisible(prunePaths))
    }
    invisible(lapply(
        X = prunePaths,
        FUN = function(path) {
            shell(
                command = "aws",
                args = append(
                    x = c(
                        "--profile", dict[["profile"]],
                        "s3", "rm"
                    ),
                    values = path
                ),
                print = TRUE
            )
        }
    ))
    alertSuccess(sprintf(
        "Pruning of binaries in {.url %s} was successful.",
        paste0("s3://", dict[["bucket"]], "/", dict[["subdir"]], "/")
    ))
    invisible(prunePaths)
}