Merge pull request #195 from NCEAS/develop

add dataset categorization helper

Author: jeanetteclark

.github/workflows/R-CMD-check.yaml

@@ -24,6 +24,9 @@ jobs:
 
       - uses: r-lib/actions/setup-pandoc@master
 
+      - uses: actions/checkout@master
+      - uses: r-lib/actions/setup-tinytex@v1
+
       - name: Cache R packages
         uses: actions/cache@v1
         if: runner.os != 'Windows'

.travis.yml

@@ -9,9 +9,11 @@ export(create_dummy_package)
 export(create_dummy_package_full)
 export(create_dummy_parent_package)
 export(create_resource_map)
+export(eml_adcad_annotation)
 export(eml_add_entity_system)
 export(eml_add_publisher)
 export(eml_associated_party)
+export(eml_categorize_dataset)
 export(eml_contact)
 export(eml_creator)
 export(eml_ecso_annotation)

NAMESPACE

@@ -644,3 +644,42 @@ eml_add_entity_system <- function(doc){
 
 
 }
+
+#' Categorize a dataset with an annotation
+#'
+#' Creates an annotation from the ADC Academic Disciplines ontology
+#' [here](https://bioportal.bioontology.org/ontologies/ADCAD/?p=classes&conceptid=root)
+#' and inserts the annotation into the EML document `doc` while retaining any existing
+#' annotations such as the sensitivity annotations. For a list of available disciplines,
+#' see link above.
+#'
+#'
+#'
+#' @param doc (emld) An EML document
+#' @param discipline (character) One or more disciplines in title case from the ADCAD ontology.
+#'
+#' @return doc (emld) An EML document with annotation added
+#' @export
+#' @examples
+#' library(EML)
+#' # read in any EML document
+#' doc <- read_eml(system.file("extdata/strix-pacific-northwest.xml", package="dataone"))
+#' # add the dataset categories
+#' doc <- eml_categorize_dataset(doc, c("Soil Science", "Ecology"))
+#'
+eml_categorize_dataset <- function(doc, discipline){
+
+  stopifnot("emld" %in% class(doc))
+
+  if (is.null(doc$dataset$id)){
+    doc$dataset$id <- gsub(":", "-", doc$packageId)
+  }
+
+  existing_anns <- doc$dataset$annotation
+
+  new_ann <- purrr::map(discipline, eml_adcad_annotation)
+
+  doc$dataset$annotation <- c(list(existing_anns), new_ann)
+
+  return(doc)
+}

R/eml.R

@@ -12,15 +12,20 @@
 read_ontology <- function(ontology_name) {
   # get the owl file from github
   if(ontology_name == "mosaic"){
-    mosaic_url <-
+    ann_url <-
       "https://raw.githubusercontent.com/DataONEorg/sem-prov-ontologies/main/MOSAiC/MOSAiC.owl"
-    mosaic <- rdflib::rdf_parse(pins::pin(mosaic_url),
+    ont <- rdflib::rdf_parse(pins::pin(ann_url),
                                 format = "rdfxml")
   } else if(ontology_name == "ecso"){
-    mosaic_url <-
+    ann_url <-
       "https://raw.githubusercontent.com/DataONEorg/sem-prov-ontologies/ECSO8-add_non-carbon_measurements/observation/ECSO8.owl"
-    mosaic <- rdflib::rdf_parse(pins::pin(mosaic_url),
+    ont <- rdflib::rdf_parse(pins::pin(ann_url),
                                 format = "rdfxml")
+  } else if (ontology_name == "ADCAD"){
+    ann_url <- "https://data.bioontology.org/ontologies/ADCAD/download?apikey=8b5b7825-538d-40e0-9e9e-5ab9274a9aeb&download_format=rdf"
+    ont <- rdflib::rdf_parse(ann_url,
+                             format = "rdfxml")
+
   }
 
 }
@@ -91,3 +96,43 @@ eml_ecso_annotation <- function(valueLabel){
                     valueURI = annotations$iri)
   )
 }
+
+#' Given a term from the ADC Academic Disciplines (ADCAD) ontology, produce the corresponding annotation
+#'
+#' Reduces the amount of copy pasting needed
+#'
+#' @param valueLabel (character) One of the disciplines found in
+#' [ADCAD](https://bioportal.bioontology.org/ontologies/OBOE/?p=classes&conceptid=root)
+#'
+#' @return list - a formatted EML annotation
+#' @export
+#'
+#' @examples eml_ecso_annotation("latitude coordinate")
+eml_adcad_annotation <- function(valueLabel){
+
+  adcad <- read_ontology("ADCAD")
+
+  query <-
+    "PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
+   PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
+
+   SELECT ?iri ?label
+   WHERE {
+     ?iri rdf:type <http://www.w3.org/2002/07/owl#Class> .
+     ?iri rdfs:label ?label .
+   }"
+
+  df <- suppressMessages(rdflib::rdf_query(adcad, query))
+
+  stopifnot(valueLabel %in% df$label)
+
+  annotations <- dplyr::filter(df, label == valueLabel)
+
+  list(
+    propertyURI = list(label = "theme",
+                       propertyURI = "http://www.w3.org/ns/dcat#theme"),
+    valueURI = list(label = annotations$label,
+                    valueURI = annotations$iri)
+  )
+}
+

R/ontology.R

@@ -368,3 +368,39 @@ test_that('Identifier systems can be added', {
   expect_equal(doc$dataset$otherEntity[[1]]$system, "https://tools.ietf.org/html/rfc4122")
   expect_equal(doc$dataset$otherEntity[[2]]$system, "https://search.dataone.org")
 })
+
+
+test_that('Datasets can be categorized', {
+
+  # two disciplines
+  doc <- read_eml("https://arcticdata.io/metacat/d1/mn/v2/object/doi%3A10.18739%2FA27D2Q84F")
+
+  doc <- eml_categorize_dataset(doc, c("Atmospheric Science", "Chemistry"))
+
+  expect_true(eml_validate(doc))
+  expect_true("Atmospheric Science" %in% eml_get_simple(doc$dataset$annotation, "label"))
+  expect_true("Chemistry" %in% eml_get_simple(doc$dataset$annotation, "label"))
+
+  # one discipline
+  doc <- read_eml("https://arcticdata.io/metacat/d1/mn/v2/object/doi%3A10.18739%2FA27D2Q84F")
+
+  doc <- eml_categorize_dataset(doc, c("Atmospheric Science"))
+
+  expect_true(eml_validate(doc))
+  expect_true("Atmospheric Science" %in% eml_get_simple(doc$dataset$annotation, "label"))
+
+  # no previous annotations with one
+
+  doc <- read_eml("https://arcticdata.io/metacat/d1/mn/v2/object/doi%3A10.18739%2FA2PV6B79W")
+  doc <- eml_categorize_dataset(doc, c("Oceanography"))
+
+  expect_true(eml_validate(doc))
+
+  # no previous annotations with one
+
+  doc <- read_eml("https://arcticdata.io/metacat/d1/mn/v2/object/doi%3A10.18739%2FA2PV6B79W")
+  doc <- eml_categorize_dataset(doc, c("Oceanography", "Mathematics"))
+
+  expect_true(eml_validate(doc))
+
+})