get_cran_marc_codes.R

library(desc)
library(httr)
library(purrr)
library(stringi)
library(methods)
library(dplyr)
library(tidyr)
pkgs <- rownames(available.packages())

if(!dir.exists("DESCRIPTIONS")) dir.create("DESCRIPTIONS")
walk(pkgs, function(p) {
    if(!file.exists(paste0("DESCRIPTIONS/", p))) {
    GET(paste0("https://raw.githubusercontent.com/cran/", p, "/master/DESCRIPTION"),
        write_disk(paste0("DESCRIPTIONS/", p)))
    }
})

pd <- data_frame(
    package = list.files("DESCRIPTIONS/"),
    role  = map(package, function(p) {
        unique(stri_extract_all_regex(
            desc_get("Author", paste0("DESCRIPTIONS/", p)), 
            "(?<=(\\[|,\\s))\\b[a-z]{3}\\b(?=(,|\\]))"))[[1]]
    }))
    

tab <- pd %>% unnest() %>% 
    filter(!is.na(role)) %>% 
    group_by(package, role) %>% 
    summarize(n= n()) %>% 
    group_by() %>% 
    count(role) %>% 
    rename(pkgs_with=nn) %>% 
    left_join(utils:::MARC_relator_db, by=c("role"="code")) %>% 
    mutate(role = if_else(role %in% utils:::MARC_relator_db_codes_used_with_R, role, paste0(role, "*"))) %>% 
    arrange(desc(pkgs_with)) %>% 
    select(-usage, -description) %>% 
    rename(code = role, `CRAN packages with`=pkgs_with) %>% 
    pander::pander_return(style="grid", justify="left")

clipr::write_clip(tab)

pd %>% unnest() %>% 
    filter(!is.na(role), !(role %in% utils:::MARC_relator_db_codes_used_with_R), role != "rev") %>% 
    pander::pander_return(style="grid", justify="left") %>% clipr::write_clip()

Another take at this 🙂

crandb <- tools::CRAN_package_db()
crandb <- janitor::clean_names(crandb)
crandb <- dplyr::filter(crandb, !is.na(authors_r))

get_roles <- function(people_string) {
  people <- eval(parse(text = people_string))
  list(format(people, include = "role", braces = list(role = c("", ""))))
}

crandb <- dplyr::rowwise(crandb)
crandb <- dplyr::mutate(
  crandb,
  role = get_roles(authors_r)
)
crandb <- dplyr::select(crandb, package, role)
crandb <- tidyr::unnest(crandb, role)
crandb <- dplyr::mutate(crandb, role = strsplit(role, split = ", "))
crandb <- tidyr::unnest(crandb, role)
marc_db <- utils:::MARC_relator_db
marc_db <- dplyr::filter(marc_db, code %in% utils:::MARC_relator_db_codes_used_with_R)
crandb <- dplyr::filter(crandb, role %in% utils:::MARC_relator_db_codes_used_with_R)
newdb <- fuzzyjoin::regex_left_join(marc_db, crandb, by = c(code = "role"))
dplyr::count(newdb, code, sort = TRUE) |> knitr::kable()

code	n
aut	19608
cre	10173
ctb	9307
cph	3757
ths	344
fnd	323
dtc	221
rev	191
trl	106
ctr	83
com	31

^{Created on 2021-07-16 by the reprex package (v2.0.0)}