build-cran-binaries/local/packages-without-binaries.R

103 lines
3.9 KiB
R

###
# Scope: List Packages for which the most recent CRAN release is not available
###
library(furrr)
library(cli)
library(dplyr)
library(data.table)
last_versions_cran <- available.packages("https://cloud.r-project.org/src/contrib")[, c("Package", "Version")]
arch <- c("amd64", "arm64")
os_ids <- c("rhel8", "rhel9", "noble", "jammy", "alpine320")
grid <- tidyr::expand_grid(arch = arch, os_ids = os_ids)
# Define the function to check for missing packages
check_package <- function(pkg_name, available_pkgs, version_cran, arch, os_id) {
if (pkg_name %in% available_pkgs$Package) {
version_devxy <- available_pkgs %>%
filter(Package == pkg_name) %>%
pull(Version)
if (is.na(version_devxy) || version_cran != version_devxy) {
# Package is missing or version mismatch
return(data.frame(pkg_name = pkg_name, arch = arch, ID = os_id))
}
} else {
# Package completely missing in devxy
return(data.frame(pkg_name = pkg_name, arch = arch, ID = os_id))
}
return(NULL) # No mismatch, no entry added
}
# Run the mapping with parallel processing for each (arch, os_id) combination
future::plan("multisession", workers = 5)
pkgs_missing <- future_map2(grid$arch, grid$os_ids, ~ {
# Prepare the URL and fetch available packages only once per (arch, os_id)
repo_url <- sprintf("https://cran.devxy.io/%s/%s/latest/src/contrib", .x, .y)
available_pkgs <- as.data.table(available.packages(contriburl = repo_url)[, c("Package", "Version")])
# Define the packages to check once
cran_pkgs <- as.data.table(last_versions_cran)
# Accumulate results in a list
pkg_entries <- list()
# Loop through each package name and check if it's missing
for (pkg_name in cran_pkgs$Package) {
version_cran <- cran_pkgs[Package == pkg_name, Version]
result <- check_package(pkg_name, available_pkgs, version_cran, .x, .y)
if (!is.null(result)) {
pkg_entries[[length(pkg_entries) + 1]] <- result
}
}
# Combine results for the current (arch, os_id)
if (length(pkg_entries) > 0) {
return(rbindlist(pkg_entries))
} else {
return(data.table(pkg_name = character(), arch = character(), ID = character()))
}
})
df_missing_pkgs <- rbindlist(pkgs_missing, fill = TRUE)
# Group and count missing packages by architecture and ID
df_summary <- df_missing_pkgs %>%
group_by(arch, ID) %>%
arrange(arch, ID) %>%
count()
# Extract and copy missing package names for each architecture and ID combination
extract_and_copy_pkgs <- function(df, arch, range = 300, id, clip = FALSE) {
pkgs <- df %>%
filter(arch == arch, ID == id) %>%
pull(pkg_name)
# Format, remove line breaks, and print as a single line
formatted_pkgs <- gsub('"', "'", capture.output(dput(pkgs[1:range])))
formatted_pkgs_one_line <- paste(formatted_pkgs, collapse = " ")
cat(formatted_pkgs_one_line, "\n", sep = "")
if (clip) {
clipr::write_clip(formatted_pkgs_one_line)
}
return(pkgs)
}
# Usage for different architecture and ID combinations
pkgs_missing_amd64_rhel8 <- extract_and_copy_pkgs(df_missing_pkgs, "amd64", "rhel8", range = 20, clip = TRUE)
pkgs_missing_amd64_rhel9 <- extract_and_copy_pkgs(df_missing_pkgs, "amd64", "rhel9", clip = TRUE)
pkgs_missing_amd64_jammy <- extract_and_copy_pkgs(df_missing_pkgs, "amd64", "jammy", clip = TRUE)
pkgs_missing_amd64_noble <- extract_and_copy_pkgs(df_missing_pkgs, "amd64", "noble", clip = TRUE)
pkgs_missing_amd64_alpine320 <- extract_and_copy_pkgs(df_missing_pkgs, "amd64", "alpine320", clip = TRUE)
pkgs_missing_arm64_rhel8 <- extract_and_copy_pkgs(df_missing_pkgs, "arm64", "rhel8", clip = TRUE)
pkgs_missing_arm64_rhel9 <- extract_and_copy_pkgs(df_missing_pkgs, "arm64", "rhel9", clip = TRUE)
pkgs_missing_arm64_jammy <- extract_and_copy_pkgs(df_missing_pkgs, "arm64", "jammy", clip = TRUE)
pkgs_missing_arm64_noble <- extract_and_copy_pkgs(df_missing_pkgs, "arm64", "noble", clip = TRUE)
pkgs_missing_arm64_alpine320 <- extract_and_copy_pkgs(df_missing_pkgs, "arm64", "alpine320", clip = TRUE)