Validated R packages

Context

The code and license is at github.com/insightsengineering/rvalidationhub-packages.

This document summarises validated R packages by R version as it was at 2024-07-12 09:59:30.52515.

It is important to note R package functionality is validated via a risk based approach, and package functionality is tested and documented against specific environments as a rolling cohort. R packages are not intended to be used outside the ‘snapshot’ they were validated within. In this repo you cannot see the environments we use, or how the package cohorts are defined temporally so cannot assume the packages will be stable on your environment and with a different cohort of package versions.

A small number of validated packages are not open sourced, usually as they are specific to our infrastructure (e.g. wrappers to simplify working against our Redshift, S3 or Snowflake instances).

We do not currently have a concept of ‘validated’ Shiny applications (we do have a business process, in the same way study code is QC’d, not ‘validated’ via a CSV process).

Output

A table is shown in the footer of this document, and a parquet file is saved in the root of this repo validated_packages.parquet.

Code

library(httr)
library(stringr)
library(dplyr)
library(glue)
library(tibble)
library(purrr)
library(tidyr)
library(gt)
library(arrow)
library(tictoc)

# variables
host <- "https://packages.roche.com/"

# get all repos
response <- httr::content(httr::GET(
  paste0(host, "__api__/repos")
))

# flatten response into a tibble
tibble_repos <- tibble(
  id = unlist(map(response, "id")),
  name = unlist(map(response, "name"))
  ) |>
  # keeping only validated
  dplyr::filter(
    str_detect(`name`,pattern = "^Validated")
  ) |>
  # new column for R version
  dplyr::mutate(
    r_version = str_extract(`name`, pattern = "\\d+\\.\\d+")
  ) |>
  # remove 3.10
  filter(r_version != "3.10")
  
# get all packages for a repo
validated_packages <- tibble()
for (i in 1:nrow(tibble_repos)) {
  # take single repo
  i_row <- slice(tibble_repos,i)
  # raw list of current packages
  response <- httr::content(httr::GET(
    glue("{host}__api__/repos/{i_row$id}/packages?_limit=1000")
  ))
  # package version live right now
  tibble_packages <- tibble(
    r_version = i_row$r_version,
    name = unlist(map(response, "name")),
    version = unlist(map(response, "version"))
  ) 
  # Check no pagination
  if (length(response) > 999) stop("Pagination not implemented: 1000+ packages in a repo.")
  # get archives
  for (j in 1:nrow(tibble_packages)) {
    
    j_row <- slice(tibble_packages,j)
    
    # get package details
    response <- httr::content(httr::GET(
      glue("{host}__api__/repos/{i_row$id}/packages/{j_row$name}")
    ))
    
    validated_packages <- j_row |>
      bind_rows(validated_packages)
    
    # archived
    if (!is.null(response$archived)){
      archived_packages <- tibble(
        r_version = i_row$r_version,
        name = unlist(map(response$archived, "name")),
        version = unlist(map(response$archived, "version"))
        ) 
      
      validated_packages <- validated_packages |>
        bind_rows(archived_packages)
    }

  }
}

# Ordering
validated_packages <- validated_packages |>
  arrange(
    r_version,
    name,
    version
  )

# Save copy
write_parquet(validated_packages, "validated_packages.parquet")

# wide table
validated_packages |>
  tidyr::pivot_wider(
    names_from = r_version,
    values_from = version,
    values_fn = list
  ) |>
  arrange(name) |>
  gt() |>
  tab_header(
    title = "Validated R packages",
    subtitle = glue("Generated on {Sys.Date()}, grouped by R version validated against")
  ) |>
  opt_interactive(
    use_search = TRUE,
    use_filters = TRUE,
    use_resizers = TRUE,
    use_highlight = TRUE,
    use_compact_mode = FALSE,
    use_text_wrapping = TRUE,
    use_page_size_select = TRUE
  )

Validated R packages

Generated on 2024-07-12, grouped by R version validated against

Execution took 144.404 seconds