In uva-bi-sdad/oss: Measuring the Cost and Value of Open Source Software

# run this instead of clicking `knit`
rmarkdown::render(here::here('./src/exploratory/code_gov/repository_domains.Rmd'),
                  output_dir = here::here('./output/code_gov'))
bad_html <- './src/exploratory/code_gov/repository_domains.html'
if (file.exists(here::here(bad_html))) file.remove(here::here(bad_html))

knitr::opts_chunk$set(echo = TRUE)
library(magrittr)
library(here)
library(tibble)
library(dplyr)
library(urltools)

options(max.print=1000000)

source(here::here('./R/file_paths.R'))

code_gov_df <- readr::read_delim(here::here(.GlobalEnv$CODE_GOV_CLEANED_FILE), delim = '\t')

print(sprintf('data location: %s', .GlobalEnv$CODE_GOV_CLEANED_FILE))

repo_url <- code_gov_df #%>%
  #select(repoID, name, agency.acronym, downloadURL, repositoryURL, permissions.licenses.name, repositoryURL_domain)

repo_github <- repo_url %>%
  filter(stringr::str_detect(repositoryURL_domain, 'github'))

repo_bitbucket <- repo_url %>%
  filter(stringr::str_detect(repositoryURL_domain, 'bitbucket'))

repo_gitlab <- repo_url %>%
  filter(stringr::str_detect(repositoryURL_domain, 'gitlab'))

non <- repo_url %>%
  filter(!repoID %in% c(repo_github$repoID, repo_bitbucket$repoID, repo_gitlab$repoID))

Domains by agency

The domain is the "base" of the URL.

All domain counts by agency

domain_count <- table(code_gov_df$repositoryURL_domain,
                      code_gov_df$agency.acronym,
                      useNA = 'always') %>%
  addmargins()
save_dir <- './data/oss/working/code_gov/domain_counts/domains_by_agency.csv'
write.csv(domain_count, here::here(save_dir))
print(domain_count, n = 99999999)

The raw unfiltered frequency table of domains by agencies can be found here: r save_dir.

The project repositoryURL's that are not github, bitbucket, or gitlab

domain_count <- table(non$repositoryURL_domain,
                      non$agency.acronym,
                      useNA = 'always') %>%
  addmargins()
save_dir <- './data/oss/working/code_gov/domain_counts/domains_by_agency_no_github_bitbucket_gitlab.csv'
write.csv(domain_count, here::here(save_dir))
print(domain_count, n = 9999)

The frequency table of the non github/bitbucket/gitlab domains by agencies can be found in r save_dir.

Licenses

If it not an OSI approved licence then we don't have to worry about the fact we cannot get the code.

Where TRUE is an OSI approved licence, and FALSE is not (including NA)

table(code_gov_df$osi_approved, useNA = 'always') %>% addmargins()

Only a 1/3 are OSI approved.

OSI by agency:

addmargins(table(code_gov_df$agency.acronym,
                 code_gov_df$osi_approved,
                 useNA = 'always'))

Licence by agency

addmargins(table(
  code_gov_df$permissions.licenses.name,
  code_gov_df$agency.acronym,
  useNA = 'always'
))

Licenses filtered by OSI

osi <- code_gov_df %>% dplyr::filter(osi_approved == TRUE)

domain_count <- table(osi$repositoryURL_domain,
                      osi$agency.acronym,
                      useNA = 'always') %>%
  addmargins()
save_dir <- './data/oss/working/code_gov/domain_counts/domains_by_agency_osi.csv'
write.csv(domain_count, here::here(save_dir))
print(domain_count, n = 9999)

The frequency table of domains by agency filtered by OSI approved repositories can be found in r save_dir.

Non "standard" domain URLs

osi_filtered_domains <- osi %>%
  filter(!repositoryURL_domain %in% c('github', 'bitbucket', 'gitlab')) %>%
  select(agency.acronym, name, repositoryURL, repositoryURL_domain, osi_approved) %>%
  arrange(agency.acronym, repositoryURL_domain)

save_dir <- './data/oss/working/code_gov/domain_urls/agency_name_repo_domain_osi.tsv'
readr::write_delim(osi_filtered_domains, here::here(save_dir), delim = '\t')

The table of agency repository name with the domain and full URL to the project of the OSI approved projects for non github/bitbucket/gitlab domans can be found in r save_dir

original_width <- getOption('width')
options(width = 999)

knitr::kable(osi_filtered_domains)

options(width = original_width)

URLs for missing licenses

missing <- code_gov_df %>% dplyr::filter(is.na(permissions.licenses.name))

addmargins(table(
  missing$repositoryURL_domain,
  useNA = 'always'
))

Almost all the NA licenses are on github.

na_licenses <- missing %>%
  select(agency.acronym, name, repositoryURL, repositoryURL_domain, osi_approved, permissions.licenses.name) %>%
  arrange(agency.acronym, repositoryURL_domain)
save_dir <- './data/oss/working/code_gov/domain_urls/agency_name_repo_domain_na.tsv'
readr::write_delim(na_licenses, here::here(save_dir), delim = '\t')

The table of projects along with the agency acronym and repositoryURL for NA licenses can be found in r save_dir

original_width <- getOption('width')
options(width = 999)

knitr::kable(na_licenses)

options(width = original_width)

uva-bi-sdad/oss documentation built on June 5, 2019, 12:37 a.m.