In uva-bi-sdad/dc.webmd.primcare: Social Data Commons: Health, Behavior, Diet: Primary Care Access Scores

knitr::opts_chunk$set(echo = TRUE)

Load dataset json template and dataset yml

dataset_metadata <- jsonlite::read_json("dataset-metadata.json")
yml_metadata <- rmarkdown::yaml_front_matter(here::here("data-raw/dataset_create.Rmd"))

Set title

dataset_metadata$datasetVersion$metadataBlocks$citation$fields[[1]]$value <- yml_metadata$dataset$citation$title

Set subtitle

dataset_metadata$datasetVersion$metadataBlocks$citation$fields[[2]]$value <- yml_metadata$dataset$citation$subtitle

Set description

dataset_metadata$datasetVersion$metadataBlocks$citation$fields[[5]]$value[[1]]$dsDescriptionValue$value <- yml_metadata$dataset$citation$dsDescriptionValue

Set subject

dataset_metadata$datasetVersion$metadataBlocks$citation$fields[[6]]$value[[1]] <- yml_metadata$dataset$citation$subject

Set production date

dataset_metadata$datasetVersion$metadataBlocks$citation$fields[[7]]$value <- yml_metadata$dataset$citation$productionDate

Set Time period covered start date

dataset_metadata$datasetVersion$metadataBlocks$citation$fields[[8]]$value[[1]]$timePeriodCoveredStart$value <- yml_metadata$dataset$citation$timePeriodCoveredStart

Set Time period covered end date

dataset_metadata$datasetVersion$metadataBlocks$citation$fields[[8]]$value[[1]]$timePeriodCoveredEnd$value <- yml_metadata$dataset$citation$timePeriodCoveredEnd

Set data sources

## Remove existing data source entries
dataset_metadata$datasetVersion$metadataBlocks$citation$fields[[9]]$value <- NULL
## Add data source entries
data_source_l <- as.list(yml_metadata$dataset$citation$dataSources)
dataset_metadata$datasetVersion$metadataBlocks$citation$fields[[9]]$value <- data_source_l

Set authors

## Remove existing author entries
dataset_metadata$datasetVersion$metadataBlocks$citation$fields[[3]]$value <- NULL
## Add author entries
num_yml_authors <- length(yml_metadata$dataset$citation$author)
for (i in 1:num_yml_authors) {
  dataset_metadata$datasetVersion$metadataBlocks$citation$fields[[3]]$value[[i]] <-
    list(
      authorName = list(
        value = yml_metadata$dataset$citation$author[[i]]$authorName,
        typeClass = "primitive",
        multiple = FALSE,
        typeName = "authorName"
      ),
      authorAffiliation = list(
        value = yml_metadata$dataset$citation$author[[i]]$authorAffiliation,
        typeClass = "primitive",
        multiple = FALSE,
        typeName = "authorAffiliation"
      )
    )
}

Set contact

## Remove existing contact entries
dataset_metadata$datasetVersion$metadataBlocks$citation$fields[[4]]$value <- NULL
## Add contact entries
num_yml_contacts <- length(yml_metadata$dataset$citation$datasetContact)
for (i in 1:num_yml_contacts) {
  dataset_metadata$datasetVersion$metadataBlocks$citation$fields[[4]]$value[[i]] <-
    list(
      datasetContactName = list(
        value = yml_metadata$dataset$citation$datasetContact[[i]]$datasetContactName,
        typeClass = "primitive",
        multiple = FALSE,
        typeName = "datasetContactName"
      ),
      datasetContactEmail = list(
        value = yml_metadata$dataset$citation$datasetContact[[i]]$datasetContactEmail,
        typeClass = "primitive",
        multiple = FALSE,
        typeName = "datasetContactEmail"
      )
    )
}

Set geographic coverage

## Remove existing geographic coverage entries
dataset_metadata$datasetVersion$metadataBlocks$geospatial$fields[[1]]$value <- NULL
## Add geographic coverage entries
nms <- names(yml_metadata$dataset$geospatial$geographicCoverage)

if ("country" %in% nms) {
  dataset_metadata$datasetVersion$metadataBlocks$geospatial$fields[[1]]$value[[1]]$country <-
    list(
      typeName = "country",
      multiple = FALSE,
      typeClass = "controlledVocabulary",
      value = yml_metadata$dataset$geospatial$geographicCoverage$country
    )
}
if ("state" %in% nms) {
  dataset_metadata$datasetVersion$metadataBlocks$geospatial$fields[[1]]$value[[1]]$state <- 
    list(
      typeName = "state",
      multiple = FALSE,
      typeClass = "primitive",
      value = yml_metadata$dataset$geospatial$geographicCoverage$state
  )
}
if ("city" %in% nms) {
  dataset_metadata$datasetVersion$metadataBlocks$geospatial$fields[[1]]$value[[1]]$city <- 
    list(
      typeName = "city",
      multiple = FALSE,
      typeClass = "primitive",
      value = yml_metadata$dataset$geospatial$geographicCoverage$city
  )
}
if ("otherGeographicCoverage" %in% nms) {
  dataset_metadata$datasetVersion$metadataBlocks$geospatial$fields[[1]]$value[[1]]$otherGeographicCoverage <- 
    list(
      typeName = "otherGeographicCoverage",
      multiple = FALSE,
      typeClass = "primitive",
      value = yml_metadata$dataset$geospatial$geographicCoverage$otherGeographicCoverage
  )
}

Set geographic unit

## Remove existing geographic unit entries
dataset_metadata$datasetVersion$metadataBlocks$geospatial$fields[[2]]$value <- NULL
## Add geographic unit entries
geo_unit_l <- as.list(yml_metadata$dataset$geospatial$geographicUnit)
dataset_metadata$datasetVersion$metadataBlocks$geospatial$fields[[2]]$value <- geo_unit_l

Create or update dataverse dataset

if (file.exists(here::here("data/dataset_doi"))) {
  dataset_doi <- readLines(here::here("data/dataset_doi"))[1]

  if (is.na(dataset_doi)) {
    ## Create dataset
    ds_info <- dataverse::create_dataset(
      dataverse = "biocomplexity",
      body = dataset_metadata,
      key = Sys.getenv("DATAVERSE_KEY"),
      server = Sys.getenv("DATAVERSE_SERVER")
    )

    ## Save dataset DOI to data
    dataset_doi <- ds_info$data$persistentId
    writeLines(dataset_doi, here::here("data/dataset_doi"))
    save(dataset_doi, file = here::here("R/sysdata.rda"))

    ## Add WebApps team as curators to the dataset

    role = list(assignee = "&explicit/5710-SDAD_WebApps", role = "curator")  
    u = paste0("https://", Sys.getenv("DATAVERSE_SERVER"), 
               "/api/datasets/:persistentId/assignments?persistentId=", 
               readLines(here::here("data/dataset_doi"))[1])  
    r <- httr::POST(u, httr::add_headers("X-Dataverse-key" = Sys.getenv("DATAVERSE_KEY"), 
                                         "Content-Type" = "application/json"), 
                    body = role, encode = 'json') 
    httr::stop_for_status(r)
    httr::content(r, "text")

  } else {

    # remove datasetVersion layer from JSON
    updated_metadata <- dataset_metadata$datasetVersion

    dataverse::update_dataset(
      dataset = dataset_doi,
      body = updated_metadata,
      key = Sys.getenv("DATAVERSE_KEY"),
      server = Sys.getenv("DATAVERSE_SERVER")
    )
  }
}

uva-bi-sdad/dc.webmd.primcare documentation built on June 18, 2022, 7 a.m.