Create or update a local MRAN snapshot for a list of packages

create-local-MRAN-snapshot.R

#!/usr/bin/env r

library(miniCRAN)
library(data.table)
library(pander)
library(logger)
library(magrittr)
library(docopt)

## #############################################################################
## config

"Make local MRAN mirror for a given date in the current working directory's {date} folder

Usage: create-local-MRAN-snapshot.R [options] <packages>...

Options:
--date=d    YYYY-MM-DD pointer to the MRAN snapshot date [default: 2019-01-01]
--force     re-download everything even if the {date} folder already exists
" %>% docopt -> opts

## #############################################################################
## prep folder

SNAPSHOT_DATE <- opts$date
SNAPSHOT_URL  <- file.path('https://mran.microsoft.com/snapshot', SNAPSHOT_DATE)
log_info('MRAN mirror: ', SNAPSHOT_URL)

if (opts$force) {
    log_info(
        'Killing local snapshot {SNAPSHOT_DATE} ',
        'with {length(list.files(SNAPSHOT_DATE, recursive = TRUE))} files')
    unlink(SNAPSHOT_DATE, recursive = TRUE)
}

## #############################################################################
## check on prior runs

dir.create(SNAPSHOT_DATE, showWarnings = FALSE)
files <- list.files(SNAPSHOT_DATE, recursive = TRUE, full.names = TRUE)
files <- cbind(path = files, file = basename(files), rbindlist(lapply(files, file.info)))

already_downloaded_packages <- sub('^([^_]*).*$', '\\1', files$file)
already_downloaded_packages <- already_downloaded_packages[!grepl('^PACKAGES', already_downloaded_packages)]

log_info('{length(already_downloaded_packages)} already downloaded R packages')
pandoc.list(already_downloaded_packages)

## #############################################################################
## list new packages

## get the list of explicitly required packages
log_info('{length(opts$packages)} R packages listed explicitly:')
pandoc.list(opts$packages)

## add all required R packages as per dependency graph
packages <- pkgDep(opts$packages, repos = SNAPSHOT_URL, type = "source", suggests = FALSE)
log_info('{length(packages)} R packages identified after looking up dependencies:')
pander(
    data.table(available.packages(repos = SNAPSHOT_URL))[Package %in% packages, .(Package, Version)],
    style = 'simple', justify = 'right')

packages <- setdiff(packages, already_downloaded_packages)
log_info('{length(packages)} R packages to be downloaded and added to the local repo:')
pandoc.list(packages)

if (length(packages) == 0) {
    log_info('Nothing to download now, exiting')
    quit(save = 'no', status = 0L)
}

## #############################################################################
## download new packages

if (length(already_downloaded_packages) > 0) {
    ## repo already exists, we just need to add new stuff
    addPackage(packages, path = SNAPSHOT_DATE, repos = SNAPSHOT_URL, deps = FALSE, type = 'source')
} else {
    ## it's a new repo
    makeRepo(packages, path = SNAPSHOT_DATE, repos = SNAPSHOT_URL, type = 'source')
}

files <- list.files(SNAPSHOT_DATE, recursive = TRUE, full.names = TRUE)
files <- cbind(path = files, file = basename(files), rbindlist(lapply(files, file.info)))
log_info('Overall {nrow(files)} files downloaded so far:')
pander(files[, .(file, size)], style = 'simple', justify = 'right')