count package authors from DESCRIPTION

count_package_authors.R

# See https://github.com/ropenscilabs/annual-report-help/issues/4

# GitHub GRAPHQL REST API v4 doesn't support any information about the changes themselves and the files that were changed
# https://git.521000.bestmunity/t5/GitHub-API-Development-and/Get-a-repository-s-commits-along-with-changed-patches-and-the/td-p/14436

# ToDo
#   - add pagination -> https://stackoverflow.com/a/48117383
#   - set repo name in 'qry$query' programmatically
#   - set 5secs time delay for the request  

library(lubridate)
library(ghql)
library(jsonlite)
library(desc) # remotes::install_github("r-lib/desc")
library(httr)

repo <- 'taxize'

token <- Sys.getenv("GITHUB_GRAPHQL_TOKEN")

cli <- GraphqlClient$new(
  url = "https://api.github.com/graphql",
  headers = add_headers(Authorization = paste0("Bearer ", token))
)

cli$load_schema()
qry <- Query$new()

# from https://stackoverflow.com/a/47502868
qry$query('getauthors', '{
  repository(owner: "ropensci", name: "taxize") {
    defaultBranchRef{
      target {
        ...on Commit{
          history(first:100,path: "DESCRIPTION"){
            nodes {
              author {
                name
                email
              }
              message
              oid
              committedDate
            }
          }
        }
      }
    }
  }
}')

qyr_result <- cli$exec(qry$queries$getauthors)

get_authors <- function(qyr_result, repo) {

  qyr_result <- fromJSON(qyr_result)
  
  res <- qyr_result$data$repository$defaultBranchRef$target$history$nodes
  
  author_name  <- res$author$name
  author_email <- res$author$email
  commit_time  <- lubridate::ymd_hms(res$committedDate)
  commit_url   <- paste0('https://raw.githubusercontent.com/ropensci/', repo, '/', res$oid, '/DESCRIPTION')
  
  df <- data.frame(reposititory = repo, 
                   author_name, 
                   author_email, 
                   commit_time, 
                   commit_url, stringsAsFactors = FALSE)
  
  df$n_authors <- purrr::map_int(df$commit_url, ~{ 
    GET(.x, write_disk(tf <- tempfile()))
    file_DESCRIPTION <- desc::desc(tf)
    n_authors <- length(file_DESCRIPTION$get_authors())
    unlink(tf)
    n_authors}) 

  df
  
}

fin <- get_authors(qyr_result, repo)
fin

n_authors.csv