Merge pull request #161 from RobLBaker/main

add test_content_units

DESCRIPTION

@@ -33,6 +33,7 @@ Imports:
     httr,
     jsonlite,
     QCkit,
+    EMLeditor,
     lifecycle
 Remotes: 
     https://github.com/NCEAS/arcticdatautils,

NAMESPACE

@@ -7,6 +7,7 @@ export(load_data)
 export(load_metadata)
 export(run_congruence_checks)
 export(test_attribute_defs)
+export(test_content_units)
 export(test_creator)
 export(test_cui_dissemination)
 export(test_datatable_urls)

NEWS.md

@@ -1,4 +1,7 @@
 # DPchecker 1.0.1 (under development)
+## 2025-03-07
+  * Add `test_content_units()` function to test for the presence of NPS content unit links. Add `test_content_units()` function to list of functions run by `run_congruence_checks()`. Add unit tests for `test_content_units()`. Add documentation about `test_content_units()` to the Articles.
+  * Add EMLeditor as a dependency to support unit tests for `test_content_units()`.
 ## 2025-02-25
   * Update `CONTRIBUTING.md`
 ## 2025-02-22

R/optional_eml_elements.R

@@ -496,7 +496,7 @@ test_public_points <- function(metadata = load_metadata(directory)){
         cli::cli_inform(c("v" = "CUI is set to PUBLIC and all GPS coordinates will be publicly available."))
       }
       else {
-        # if CUI is not public, warn the user that GPS coordiantes will be public.
+        # if CUI is not public, warn the user that GPS coordinates will be public.
         cli::cli_warn(c("!" = "CUI is not set to PUBLIC. GPS coordinates detected in metadata will be publicly available. Are you sure?"))
       }
     }
@@ -545,7 +545,7 @@ test_project <- function (metadata = load_metadata(directory)) {
   proj_test <- unlist(proj)
   DS_proj <- "id" %in% names(proj_test)
   if (!DS_proj) {
-    msg <- paste0("No project associated with the metadata. ",
+    msg <- paste0("No DataStore project associated with the metadata. ",
                   "To add a DataStore project, use ",
                   "{.fun EMLeditor::set_project}.")
     cli::cli_warn(c("!" = msg))
@@ -554,5 +554,70 @@ test_project <- function (metadata = load_metadata(directory)) {
 
   msg <- "The metadata contains at least one DataStore Project reference."
   cli::cli_inform(c("v" = msg))
+  return(invisible(metadata))
 }
 
+#' Test for content unit links
+#'
+#' @inheritParams test_pub_date
+#'
+#' @return invisible(metadata)
+#' @export
+#'
+#' @examples
+#'  \dontrun{
+#' test_content_units()
+#' }
+test_content_units <- function(metadata = load_metadata(directory)) {
+  #check whether EML is schema valid
+  is_eml(metadata)
+
+  #get geographic coverage from metadata
+  geo_cov <- metadata[["dataset"]][["coverage"]][["geographicCoverage"]]
+
+  #drop `@context` item from geo_cov
+  geo_cov$`@context` <- NULL
+
+  # If there's only geographic coverage element, geo_cov ends up with one less level of nesting. Re-nest it so that the rest of the code works consistently
+  if ("geographicDescription" %in% names(geo_cov)) {
+    geo_cov <- list(geo_cov)
+  }
+
+  #test for existence of geography; if absent warn and suggest solution
+  if (is.null(geo_cov)) {
+    msg1 <- "Metadata does not contain park content unit links."
+    msg <- "to add content unit links."
+    cli::cli_warn(
+      c("!" = "{msg1}, Use {.fn EMLeditor::set_content_units} {msg}"))
+      }
+
+  #if geography is present:
+  else {
+    units <- FALSE
+    for (i in 1:length(seq_along(geo_cov))) {
+      #test for content unit links:
+      if (grepl("NPS Content Unit Link:",
+                geo_cov[[i]]$geographicDescription)) {
+        units <- TRUE
+      }
+      # exit the for-loop after the first instance of a content unit:
+      if (units == TRUE) {
+        break
+      }
+    }
+    # if content units are present, pass the test
+    if (units) {
+      cli::cli_inform(c("v" = "Metadata contains NPS content unit links."))
+    }
+    #if content units are not present, fail with a warning and suggest solution
+    if (!units) {
+      msg1 <- "Metadata does not contain park content unit links."
+      msg <- "to add content unit links."
+      cli::cli_warn(
+        c("!" = "{msg1}, Use {.fn EMLeditor::set_content_units} {msg}"))
+
+    }
+  }
+
+  return(invisible(metadata))
+}

R/run_checks.R

@@ -143,6 +143,15 @@ run_congruence_checks <- function(directory = here::here(),
              warn_count <<- warn_count + 1
              cli::cli_bullets(c(w$message, w$body))
            })
+  tryCatch(test_content_units(metadata),
+           error = function(e) {
+             err_count <<- err_count + 1
+             cli::cli_bullets(c(e$message, e$body))
+           },
+           warning = function(w) {
+             warn_count <<- warn_count + 1
+             cli::cli_bullets(c(w$message, w$body))
+           })
   tryCatch(test_doi(metadata),
            error = function(e) {
              err_count <<- err_count + 1

docs/pkgdown.yml

@@ -3,4 +3,4 @@ pkgdown: 2.1.1
 pkgdown_sha: ~
 articles:
   DPchecker: DPchecker.html
-last_built: 2025-02-25T21:14Z
+last_built: 2025-03-07T18:37Z

docs/sitemap.xml

@@ -19,6 +19,7 @@
 <url><loc>/reference/run_congruence_checks.html</loc></url>
 <url><loc>/reference/test_attribute_defs.html</loc></url>
 <url><loc>/reference/test_by_for_nps.html</loc></url>
+<url><loc>/reference/test_content_units.html</loc></url>
 <url><loc>/reference/test_creator.html</loc></url>
 <url><loc>/reference/test_cui_dissemination.html</loc></url>
 <url><loc>/reference/test_datatable_urls.html</loc></url>

tests/testthat/test-optional_eml_elements.R

@@ -88,3 +88,25 @@ test_that("test_project warns when no project is present",
                            "No project associated with the metadata.*")
             }
           )
+
+# ---- test_content_units
+test_that("test_content_units warns when no content units links are present",
+          {
+            expect_warning(test_content_units(metadata = bicy_meta),
+                           "Metadata does not contain park content unit*")
+})
+
+test_that("test_content_units passes when content unit links are present",
+          {
+            meta2 <- EMLeditor::set_content_units(bicy_meta, "ROMO")
+            expect_message(test_content_units(meta2),
+                           "Metadata contains NPS content unit links.")
+          })
+
+test_that("test_content_units warns when no geography present",
+          {
+            meta <- bicy_meta
+            meta[["dataset"]][["coverage"]][["geographicCoverage"]] <- NULL
+            expect_warning(test_content_units(metadata = meta),
+                           "Metadata does not contain park content unit links.*")
+          })

vignettes/DPchecker.Rmd

@@ -86,6 +86,7 @@ These tests determine whether the metadata is schema valid and adheres to some r
 * Metadata indicates that data files do not have footers ([`test_footer()`](../reference/test_footer.html))
 * Metadata contains taxonomic coverage element ([`test_taxonomic_cov()`](../reference/test_taxonomic_cov.html))
 * Metadata contains geographic coverage element ([`test_geographic_cov()`](../reference/test_geographic_cov.html))
+* Metadata contain NPS content unit links ([test_content_units()](../reference/test_content_units.html))
 * Metadata contains a Digital Object Identifier (DOI) ([`test_doi()`](../reference/test_doi.html))
 * Metadata DOI is properly formatted ([`test_doi_format()`](../reference/test_doi_format.html))
 * Metadata contains URLs for each data table  ([test_datatable_urls](../reference/test_datatable_urls.html))