feat: use correct property for labelling HAL, let users enter property

NAMESPACE

@@ -22,6 +22,7 @@ export(spq_arrange)
 export(spq_assemble)
 export(spq_control_request)
 export(spq_count)
+export(spq_endpoint_info)
 export(spq_filter)
 export(spq_group_by)
 export(spq_head)

R/build_sparql.R

@@ -13,7 +13,7 @@
 #'   cat()
 spq_assemble = function(.query, strict = TRUE) {
 
-  endpoint = .query[["endpoint"]]
+  endpoint = .query[["endpoint_info"]][["endpoint_url"]]
 
   .query = spq_prefix(.query, auto = TRUE, prefixes = NULL)
 

R/data.R

@@ -20,7 +20,7 @@ wikidata_url <- function() {
 #' \describe{
 #'   \item{name}{the abbreviated name of the SPARQL endpoint}
 #'   \item{url}{the full address of the SPARQL endpoint}
-#'   ...
+#'   \item{label_property}{the property used for labelling}
 #' }
 "usual_endpoints"
 

R/spq_endpoint_info.R

@@ -0,0 +1,23 @@
+#' Create the endpoint info object for `spq_init()`
+#'
+#' @param label_property Property used by the endpoint for labelling.
+#'
+#' @return A list to be used in `spq_init()`'s `endpoint_info` argument.
+#' @export
+#'
+#' @examples
+#' spq_endpoint_info(label_property = "skos:preflabel")
+spq_endpoint_info <- function(label_property = "rdfs:prefLabel") {
+
+  # TODO check property more
+  if (!is.character(label_property)) {
+    cli::cli_abort("Must provide a character as {.arg label_property}.")
+  }
+
+  structure(
+    list(
+      label_property = label_property
+    ),
+    class = "glitter_endpoint_info"
+  )
+}

R/spq_init.R

@@ -2,6 +2,9 @@
 #'
 #' @param endpoint Endpoint, either name if it is in `usual_endpoints`,
 #' or an URL
+#' @param endpoint_info Do not use for an usual endpoint in `usual_endpoints`!
+#' Information about
+#' the endpoint
 #' @param request_control An object as returned by [`spq_control_request()`]
 #'
 #' @return A query object
@@ -28,22 +31,35 @@ spq_init = function(
       max_seconds = getOption("glitter.max_seconds", 120L),
       timeout = getOption("glitter.timeout", 1000L),
       request_type = c("url", "body-form")
+    ),
+    endpoint_info = spq_endpoint_info(
+      label_property = "rdfs:label"
     )
   ) {
  if (!inherits(request_control, "glitter_request_control")) {
     cli::cli_abort("{.arg request_control} must be created by {.fun spq_control_request}.")
+ }
+ if (!inherits(endpoint_info, "glitter_endpoint_info")) {
+    cli::cli_abort("{.arg endpoint_info} must be created by {.fun spq_endpoint_info}.")
   }
 
   # if endpoint passed as name, get url
   endpoint = tolower(endpoint)
   usual_endpoint_info = usual_endpoints %>%
     dplyr::filter(.data$name == endpoint)
-  endpoint = if (nrow(usual_endpoint_info) > 0) {
-    dplyr::pull(usual_endpoint_info, .data$url)
+  if (nrow(usual_endpoint_info) > 0) {
+    endpoint = dplyr::pull(usual_endpoint_info, .data$url)
+    label_property = dplyr::pull(usual_endpoint_info, .data$label_property)
   } else {
     endpoint
+    label_property = NULL
   }
 
+  endpoint_info = list(
+    endpoint_url = endpoint,
+    label_property = label_property %||% endpoint_info[["label_property"]]
+  )
+
   query = list(
     prefixes_provided = tibble::tibble(name = NULL, url = NULL),
     prefixes_used = NULL,
@@ -56,7 +72,7 @@ spq_init = function(
     group_by = NULL,
     order_by = NULL,
     offset = NULL,
-    endpoint = endpoint,
+    endpoint_info = endpoint_info,
     request_control = request_control
   )
 

R/spq_label.R

@@ -35,6 +35,10 @@ spq_label <- function(.query,
                       .required = FALSE,
                       .languages = getOption("glitter.lang", "en$"),
                       .overwrite = FALSE) {
+
+  label_property <- .query[["endpoint_info"]][["label_property"]] %||%
+    "rdfs:label"
+
   vars = purrr::map_chr(rlang::enquos(...), spq_treat_argument)
 
   if (!is.null(.languages)) .languages = tolower(.languages)
@@ -56,14 +60,14 @@ spq_label <- function(.query,
       if (.required) {
        q = spq_add(
         query,
-        sprintf("%s rdfs:label %s_labell", x, x),
+        sprintf("%s %s %s_labell", x, label_property, x),
         .required = .required
       )
        q = spq_filter(q, spq(filter))
       } else {
       q = spq_add(
         query,
-        sprintf("%s rdfs:label %s_labell", x, x),
+        sprintf("%s %s %s_labell", x,label_property, x),
         .required = .required,
         .filter = filter
       )

R/spq_perform.R

@@ -34,7 +34,7 @@ spq_perform = function(.query,
        "spq_init(endpoint)"
     )
   } else {
-    endpoint = .query[["endpoint"]]
+    endpoint = .query[["endpoint_info"]][["endpoint_url"]]
   }
 
 

data-raw/create_usual_endpoints.R

@@ -1,14 +1,2 @@
-endpoints=tibble::tibble(name=c("wikidata",
-                                "dbpedia",
-                                "databnf",
-                                "isidore",
-                                "hal",
-                                "symogih"),
-                         url=c("https://query.wikidata.org/",
-                               "https://dbpedia.org/sparql",
-                               "https://data.bnf.fr/sparql",
-                               "https://isidore.science/sparql",
-                               "http://sparql.archives-ouvertes.fr/sparql",
-                               "http://bhp-publi.ish-lyon.cnrs.fr:8888/sparql"))
 usual_endpoints=readr::read_csv("data-raw/usual_endpoints.csv")
 usethis::use_data(usual_endpoints,overwrite=TRUE)

data-raw/usual_endpoints.csv

@@ -1,6 +1,6 @@
-name,url
-wikidata,https://query.wikidata.org/
-dbpedia,https://dbpedia.org/sparql
-databnf,https://data.bnf.fr/sparql
-isidore,https://isidore.science/sparql
-hal,http://sparql.archives-ouvertes.fr/sparql
+name,url,label_property
+wikidata,https://query.wikidata.org/,rdfs:label
+dbpedia,https://dbpedia.org/sparql,rdfs:label
+databnf,https://data.bnf.fr/sparql,rdfs:label
+isidore,https://isidore.science/sparql,rdfs:label
+hal,http://sparql.archives-ouvertes.fr/sparql,skos:prefLabel

tests/testthat/_snaps/spq_label.md

@@ -60,6 +60,30 @@
       }
       
 
+# spq_label() for not rdfs:label
+
+    Code
+      spq_init(endpoint = "hal") %>% spq_add(
+        "haldoc:inria-00362381 dcterms:hasVersion ?version") %>% spq_add(
+        "?version dcterms:type ?type") %>% spq_label(type)
+    Output
+      PREFIX dcterms: <http://purl.org/dc/terms/>
+      PREFIX skos: <http://www.w3.org/2004/02/skos/core#>
+      PREFIX haldoc: <https://data.archives-ouvertes.fr/document/>
+      SELECT ?type (COALESCE(?type_labell,'') AS ?type_label) ?version
+      WHERE {
+      
+      haldoc:inria-00362381 dcterms:hasVersion ?version.
+      ?version dcterms:type ?type.
+      OPTIONAL {
+      	?type skos:prefLabel ?type_labell.
+      	FILTER(lang(?type_labell) IN ('en'))
+      }
+      
+      
+      }
+      
+
 # spq_label() .overwrite
 
     Code

tests/testthat/test-spq_label.R

@@ -35,6 +35,15 @@ test_that("spq_label() works", {
   )
 })
 
+test_that("spq_label() for not rdfs:label", {
+  expect_snapshot(
+    spq_init(endpoint = "hal") %>%
+      spq_add("haldoc:inria-00362381 dcterms:hasVersion ?version") %>%
+      spq_add("?version dcterms:type ?type") %>%
+      spq_label(type)
+  )
+})
+
 test_that("spq_label() .overwrite", {
 
   expect_snapshot(

vignettes/articles/glitter_for_hal.Rmd

@@ -85,7 +85,7 @@ On cherche les types associés aux versions de documents. Ces types sont associ
 query_docType=spq_init(endpoint = "hal") %>% 
   spq_add("haldoc:inria-00362381 dcterms:hasVersion ?version") %>% # Ce doc a des versions ?version
   spq_add("?version dcterms:type ?type") %>%                       # ?version est un document de type ?type
-  spq_add("?type skos:prefLabel ?label")                           # ?type a pour étiquette ?label
+  spq_label(type)                           # ?type a pour étiquette ?label
 
 tib_docType=spq_perform(query_docType)