In ropensci/codemetar: Generate 'CodeMeta' Metadata for R Packages

library("gh")
library("jsonlite")
library("jsonld")
library("codemetar")
library("magrittr")
library("readr")
library("dplyr")
library("purrr")

crosswalk <- function(x,
                      schema,
                      codemeta_context = 
             "https://doi.org/10.5063/schema/codemeta-2.0"
                      ){

  df <- crosswalk_table(schema)
  context <- crosswalk_context(df, codemeta_context)
  crosswalk_transform(x, 
                      crosswalk_context = context,
                      codemeta_context = codemeta_context)

}

crosswalk_table <- function(schema){
  df <-
    read_csv("https://github.com/codemeta/codemeta/raw/master/crosswalk.csv",
             col_types = cols(.default = "c"))
  df <- df[c("Property", schema)]
  df[!is.na(df[[schema]]),] ## stats::na.omit
}

crosswalk_context <- 
  function(df, codemeta_context = 
             "https://doi.org/10.5063/schema/codemeta-2.0"){

  context <- jsonlite::read_json(codemeta_context)
  context[[1]][["id"]] <- NULL ## avoid collisions with @id
  properties <- names(context[[1]])  

  ## replace property names according to crosswalk data.frame
  for(i in 1:dim(df)[1]){
   properties[properties == df[i, 1][[1]] ]  <- df[i,2]
  }
  names(context[[1]]) <- properties
  context
}

crosswalk_transform <- function(x, 
                                crosswalk_context = NULL,
                                codemeta_context = 
             "https://doi.org/10.5063/schema/codemeta-2.0"){

  ## FIXME consider case of x is `json` class, and x is file-path / url

  if(is.null(x$`@context`) && is.null(crosswalk_context)){
    stop("crosswalk_context required if input json has no '@context' property")
  }
  if(is.null(x$`@context`))
    x$`@context` <- crosswalk_context[[1]]

  y <- toJSON(x, auto_unbox = TRUE, pretty = TRUE)
  y <- jsonld_expand(y)
  y <- jsonld_compact(y, context = codemeta_context)
  y
}

Example:

owner <- "ropensci"
repo <- "EML"
r <- gh("/repos/:owner/:repo", owner = owner, repo = repo)
crosswalk(r, "GitHub")

Add Contributor information. Get contributor list, then contributor names

contribs <- gh("/repos/:owner/:repo/contributors", owner = owner, repo = repo)

author_threshold <- 2
authors <- contribs %>% map_int("contributions") >= author_threshold

## Sort authors by contribution number.  Seems contribs already does this?
contribs[authors] %>%
  map_chr("login") %>%
  map(function(user) gh("/users/:user", user = user)) %>%
  map_chr("name")

Other mappable info: user:company, user:blog, user:email (API doesn't return email)

gh("/users/:user", user = "cboettig")

Use the preview version of the API to get license information

r <- gh("/repos/:owner/:repo", owner = owner, repo = repo, .send_headers = c(ACCEPT = "application/vnd.github.drax-preview+json"))

ropensci/codemetar documentation built on Feb. 24, 2024, 12:34 a.m.