functions to use dataverse API bindings in R to download repositories

DataverseDownload.R

library(dataverse)
library(stringr)

# add credentials (required for large jobs or gated files)

#Sys.setenv("DATAVERSE_SERVER" = "dataverse.harvard.edu")
#Sys.setenv("DATAVERSE_KEY" = "examplekey12345")

######################################################################
# download individual file from dataverse doi
dl_file = function(fn, dirn, doi){
      cat(paste0("Downloading ", fn, "\n"))
      f = try(get_file(fn, doi, orginal = TRUE))              # download to binary
      # failure state - 404 or 403 errors
      if(class(f) == "try-error"){                  
          cat(paste0(fn, " failed. \n"))
          return(NULL)
      } else { # write binary to file
        writeBin(f, file.path(dirn, fn))
        cat(paste0(fn, " downloaded! \n"))
        return(TRUE)
      }
}

# function to mirror entire dataverse doi
dv_download = function(doi, dirn, pat = ""){
    # create folder if it doesn't exist
    dir.create(dirn, showWarnings = FALSE)
    # get metadata
    dataset = get_dataset(doi)
    # list of files in repo
    fl      = dataset$files$label
    # subset to pattern
    fl = str_subset(fl, pat)
    # download all files in repo
    for(f in fl) dl_file(f, dirn, doi)
}

######################################################################

dv_download("doi:10.7910/DVN/5TBFXL", 'test')