feat: Add dataset language (#1133)

* Simplify structure * Fix Iconclass language * Add API test
netwerk-digitaal-erfgoed · Oct 11, 2023 · 3e53bc0 · 3e53bc0
1 parent 0d8f4ae
commit 3e53bc0
Show file tree

Hide file tree

Showing 50 changed files with 66 additions and 11 deletions.
diff --git a/jest.config.js b/jest.config.js
@@ -11,9 +11,9 @@ export default {
   ],
   coverageThreshold: {
     global: {
-      lines: 90.87,
-      statements: 90.87,
-      branches: 95.86,
+      lines: 90.89,
+      statements: 90.89,
+      branches: 95.89,
       functions: 91.52,
     },
   },

diff --git a/packages/network-of-terms-catalog/catalog/datasets/aat-materials.jsonld b/packages/network-of-terms-catalog/catalog/datasets/aat-materials.jsonld
@@ -28,6 +28,7 @@
       "@value": "Selectie van AAT termen voor het beschrijven van materialen in architectuur-, kunst- en cultuurhistorische collecties"
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "http://vocab.getty.edu/aat/sparql/materials",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/aat-processes-and-techniques.jsonld b/packages/network-of-terms-catalog/catalog/datasets/aat-processes-and-techniques.jsonld
@@ -28,6 +28,7 @@
       "@value": "Selectie van AAT termen voor het beschrijven van processen en technieken in architectuur-, kunst- en cultuurhistorische collecties"
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "http://vocab.getty.edu/aat/sparql/processes-and-techniques",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/aat-styles-and-periods.jsonld b/packages/network-of-terms-catalog/catalog/datasets/aat-styles-and-periods.jsonld
@@ -28,6 +28,7 @@
       "@value": "Selectie van AAT termen voor het beschrijven van stijlen en periodes in architectuur-, kunst- en cultuurhistorische collecties"
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "http://vocab.getty.edu/aat/sparql/styles-and-periods",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/aat.jsonld b/packages/network-of-terms-catalog/catalog/datasets/aat.jsonld
@@ -28,6 +28,7 @@
       "@value": "Onderwerpen voor het beschrijven van architectuur-, kunst- en cultuurhistorische collecties"
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "http://vocab.getty.edu/aat/sparql",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/abr.jsonld b/packages/network-of-terms-catalog/catalog/datasets/abr.jsonld
@@ -28,6 +28,7 @@
       "@value": "Onderwerpen voor het beschrijven van onderzoeken, vondsten en sporen"
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "https://data.cultureelerfgoed.nl/PoolParty/sparql/term/id/abr",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/adamlink-straten.jsonld b/packages/network-of-terms-catalog/catalog/datasets/adamlink-straten.jsonld
@@ -22,6 +22,7 @@
       "@value": "Straten in Amsterdam"
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "https://druid.datalegend.net/AdamNet/Geography/sparql#streets",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/brabantse-gebouwen.jsonld b/packages/network-of-terms-catalog/catalog/datasets/brabantse-gebouwen.jsonld
@@ -22,6 +22,7 @@
       "@value": "Gebouwen in de provincie Noord-Brabant, vooralsnog met een religieuze functie, zoals kloosters"
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "https://data.brabantcloud.nl/gebouwen/query/",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/brinkman.jsonld b/packages/network-of-terms-catalog/catalog/datasets/brinkman.jsonld
@@ -28,6 +28,7 @@
       "@value": "Onderwerpen die de Koninklijke Bibliotheek aan publicaties toekent"
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "http://data.bibliotheken.nl/thes/brinkman/sparql",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/cht-materials.jsonld b/packages/network-of-terms-catalog/catalog/datasets/cht-materials.jsonld
@@ -28,6 +28,7 @@
       "@value": "Onderwerpen voor het beschrijven van materiaal kenmerken van cultureel erfgoed objecten."
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "https://data.cultureelerfgoed.nl/PoolParty/sparql/term/id/cht/materials",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/cht-styles-and-periods.jsonld b/packages/network-of-terms-catalog/catalog/datasets/cht-styles-and-periods.jsonld
@@ -28,6 +28,7 @@
       "@value": "Onderwerpen voor het beschrijven van stijl en periode kenmerken van cultureel erfgoed objecten."
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "https://data.cultureelerfgoed.nl/PoolParty/sparql/term/id/cht/styles-and-periods",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/cht.jsonld b/packages/network-of-terms-catalog/catalog/datasets/cht.jsonld
@@ -28,6 +28,7 @@
       "@value": "Onderwerpen voor het beschrijven van cultureel erfgoed"
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "https://data.cultureelerfgoed.nl/PoolParty/sparql/term/id/cht",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/eurovoc.jsonld b/packages/network-of-terms-catalog/catalog/datasets/eurovoc.jsonld
@@ -22,6 +22,7 @@
       "@value": "Onderwerpen over alle gebieden waarop de Europese Unie actief is. De nadruk ligt op de parlementaire activiteiten van de EU"
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "http://publications.europa.eu/webapi/rdf/sparql#eurovoc",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/geonames.jsonld b/packages/network-of-terms-catalog/catalog/datasets/geonames.jsonld
@@ -22,6 +22,7 @@
       "@value": "Selectie van geografische namen zoals plaatsen, administratieve eenheden (gemeentes, provincies) en waterlichamen (rivieren, beken, meren etc.)"
     }
   ],
+  "inLanguage": "en",
   "distribution": [
     {
       "@id": "https://demo.netwerkdigitaalerfgoed.nl/geonames",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/goudatijdmachine-straten.jsonld b/packages/network-of-terms-catalog/catalog/datasets/goudatijdmachine-straten.jsonld
@@ -25,6 +25,7 @@
       "@value": "Straten in Gouda"
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "https://www.goudatijdmachine.nl/id/straten#streets",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/gtaa-classificatie.jsonld b/packages/network-of-terms-catalog/catalog/datasets/gtaa-classificatie.jsonld
@@ -22,6 +22,7 @@
       "@value": "Classificaties voor het beschrijven van audiovisueel materiaal"
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "https://data.beeldengeluid.nl/id/datadownload/0027",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/gtaa-genres.jsonld b/packages/network-of-terms-catalog/catalog/datasets/gtaa-genres.jsonld
@@ -22,6 +22,7 @@
       "@value": "Genres voor het beschrijven van audiovisueel materiaal"
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "https://data.beeldengeluid.nl/id/datadownload/0028",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/gtaa-geografische-namen.jsonld b/packages/network-of-terms-catalog/catalog/datasets/gtaa-geografische-namen.jsonld
@@ -22,6 +22,7 @@
       "@value": "Geografische namen voor het beschrijven van audiovisueel materiaal"
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "https://data.beeldengeluid.nl/id/datadownload/0029",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/gtaa-namen.jsonld b/packages/network-of-terms-catalog/catalog/datasets/gtaa-namen.jsonld
@@ -22,6 +22,7 @@
       "@value": "Diverse soorten eigennamen voor het beschrijven van audiovisueel materiaal"
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "https://data.beeldengeluid.nl/id/datadownload/0030",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/gtaa-onderwerpen-beng.jsonld b/packages/network-of-terms-catalog/catalog/datasets/gtaa-onderwerpen-beng.jsonld
@@ -22,6 +22,7 @@
       "@value": "Onderwerpen die slechts voor shots/clips/geluiden worden toegepast voor het beschrijven van audiovisueel materiaal"
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "https://data.beeldengeluid.nl/id/datadownload/0032",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/gtaa-onderwerpen.jsonld b/packages/network-of-terms-catalog/catalog/datasets/gtaa-onderwerpen.jsonld
@@ -22,6 +22,7 @@
       "@value": "Onderwerpen voor het beschrijven van audiovisueel materiaal"
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "https://data.beeldengeluid.nl/id/datadownload/0031",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/gtaa-persoonsnamen.jsonld b/packages/network-of-terms-catalog/catalog/datasets/gtaa-persoonsnamen.jsonld
@@ -22,6 +22,7 @@
       "@value": "Personen voor het beschrijven van audiovisueel materiaal"
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "https://data.beeldengeluid.nl/id/datadownload/0026",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/homosaurus.jsonld b/packages/network-of-terms-catalog/catalog/datasets/homosaurus.jsonld
@@ -22,6 +22,7 @@
       "@value": "Termen voor het beschrijven van LHBTIQ (Lesbisch/Homo/Biseksueel/Transgender/Intersekse/Queer)-publicaties en -erfgoed"
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "https://data.ihlia.nl/PoolParty/sparql/homosaurus",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/iconclass.jsonld b/packages/network-of-terms-catalog/catalog/datasets/iconclass.jsonld
@@ -22,6 +22,7 @@
       "@value": "Termen voor het beschrijven van de inhoud van afbeeldingen"
     }
   ],
+  "inLanguage": "en",
   "distribution": [
     {
       "@id": "https://iconclass.org/sparql",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/ied.jsonld b/packages/network-of-terms-catalog/catalog/datasets/ied.jsonld
@@ -28,6 +28,7 @@
       "@value": "Termen voor het beschrijven van collecties uit de periode 1930–1970 rondom voormalig Nederlands-Indië, onafhankelijk Indonesië en de postkoloniale migratie van personen naar Nederland"
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "https://digitaalerfgoed.poolparty.biz/PoolParty/sparql/ied",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/koloniaal-verleden.jsonld b/packages/network-of-terms-catalog/catalog/datasets/koloniaal-verleden.jsonld
@@ -22,6 +22,7 @@
       "@value": "Termen voor het beschrijven van museale collecties over het koloniaal verleden"
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "https://data.cultureelerfgoed.nl/PoolParty/sparql/koloniaalverleden",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/muziekschatten-onderwerpen.jsonld b/packages/network-of-terms-catalog/catalog/datasets/muziekschatten-onderwerpen.jsonld
@@ -22,6 +22,7 @@
       "@value": "Onderwerpen in de catalogus van de bladmuziekcollectie van Stichting Omroep Muziek"
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "https://data.muziekschatten.nl/sparql/#onderwerpen",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/muziekschatten-personen.jsonld b/packages/network-of-terms-catalog/catalog/datasets/muziekschatten-personen.jsonld
@@ -22,6 +22,7 @@
       "@value": "Personen in de catalogus van de bladmuziekcollectie van Stichting Omroep Muziek"
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "https://data.muziekschatten.nl/sparql/#personen",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/mw-genresstijlen.jsonld b/packages/network-of-terms-catalog/catalog/datasets/mw-genresstijlen.jsonld
@@ -22,6 +22,7 @@
       "@value": "Genres en stijlen in de collectie van Muziekweb"
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "https://data.muziekweb.nl/MuziekwebOrganization/Muziekweb/sparql/Muziekweb#mw-genresstijlen",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/mw-personengroepen.jsonld b/packages/network-of-terms-catalog/catalog/datasets/mw-personengroepen.jsonld
@@ -22,6 +22,7 @@
       "@value": "Artiesten, zowel personen als groepen, in de collectie van Muziekweb"
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "https://data.muziekweb.nl/MuziekwebOrganization/Muziekweb/sparql/Muziekweb#mw-personengroepen",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/nmvw.jsonld b/packages/network-of-terms-catalog/catalog/datasets/nmvw.jsonld
@@ -22,6 +22,7 @@
       "@value": "Onderwerpen verdeeld over de facetten Functie, Cultuur, Geografische herkomst, Object, Materiaal & Techniek"
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "https://data.netwerkdigitaalerfgoed.nl/NMVW/thesaurus/sparql",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/nta.jsonld b/packages/network-of-terms-catalog/catalog/datasets/nta.jsonld
@@ -28,6 +28,7 @@
       "@value": "Naams- en andere persoonsgegevens van auteurs"
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "http://data.bibliotheken.nl/thesp/sparql",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/rkdartists.jsonld b/packages/network-of-terms-catalog/catalog/datasets/rkdartists.jsonld
@@ -22,7 +22,7 @@
       "@value": "Biografische gegevens van Nederlandse en buitenlandse kunstenaars van de middeleeuwen tot heden"
     }
   ],
-  "distribution": [
+"inLanguage": "nl",  "distribution": [
     {
       "@id": "https://data.netwerkdigitaalerfgoed.nl/rkd/rkdartists/sparql",
       "@type": "DataDownload",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/stcn-drukkers.jsonld b/packages/network-of-terms-catalog/catalog/datasets/stcn-drukkers.jsonld
@@ -22,6 +22,7 @@
       "@value": "Uitgevers, een subset van Short-Title Catalogue Netherlands (STCN)"
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "http://data.bibliotheken.nl/thes/drukkers/sparql",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/uitvoeringsmedium.jsonld b/packages/network-of-terms-catalog/catalog/datasets/uitvoeringsmedium.jsonld
@@ -22,6 +22,7 @@
       "@value": "Termen voor het beschrijven van muziekinstrumenten, zangstemmen en ensembles (in enkelvoudige en meervoudige bezetting)"
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "https://data.muziekschatten.nl/sparql/#uitvoeringsmedium",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/wikidata-entities-all.jsonld b/packages/network-of-terms-catalog/catalog/datasets/wikidata-entities-all.jsonld
@@ -22,6 +22,7 @@
       "@value": "Alle entiteiten in Wikidata, zoals beroepen, films of kunstwerken"
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "https://query.wikidata.org/sparql#entities-all",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/wikidata-entities-persons.jsonld b/packages/network-of-terms-catalog/catalog/datasets/wikidata-entities-persons.jsonld
@@ -22,6 +22,7 @@
       "@value": "Personen"
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "https://query.wikidata.org/sparql#entities-persons",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/wikidata-entities-places.jsonld b/packages/network-of-terms-catalog/catalog/datasets/wikidata-entities-places.jsonld
@@ -22,6 +22,7 @@
       "@value": "Plaatsen in Nederland en België"
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "https://query.wikidata.org/sparql#entities-places",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/wikidata-entities-streets.jsonld b/packages/network-of-terms-catalog/catalog/datasets/wikidata-entities-streets.jsonld
@@ -22,6 +22,7 @@
       "@value": "Straten in Nederland"
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "https://query.wikidata.org/sparql#entities-streets",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/wo2biografie.jsonld b/packages/network-of-terms-catalog/catalog/datasets/wo2biografie.jsonld
@@ -22,6 +22,7 @@
       "@value": "Korte biografieën van personen die een sleutelrol speelden in de Tweede Wereldoorlog"
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "https://data.niod.nl/PoolParty/sparql/WO2_biografieen",

diff --git a/packages/network-of-terms-catalog/catalog/datasets/wo2thesaurus.jsonld b/packages/network-of-terms-catalog/catalog/datasets/wo2thesaurus.jsonld
@@ -22,6 +22,7 @@
       "@value": "Gebeurtenissen, plaatsen, begrippen en objecten uit de Tweede Wereldoorlog"
     }
   ],
+  "inLanguage": "nl",
   "distribution": [
     {
       "@id": "https://data.niod.nl/PoolParty/sparql/WO2_Thesaurus",

diff --git a/packages/network-of-terms-catalog/shacl/dataset.jsonld b/packages/network-of-terms-catalog/shacl/dataset.jsonld
@@ -22,6 +22,12 @@
           },
           "sh:minCount": 0
         },
+        {
+          "sh:path": {
+            "@id": "schema:inLanguage"
+          },
+          "sh:minCount": 1
+        },
         {
           "sh:path": {
             "@id": "schema:url"

diff --git a/packages/network-of-terms-catalog/src/getCatalog.ts b/packages/network-of-terms-catalog/src/getCatalog.ts
@@ -30,17 +30,17 @@ export async function getCatalog(path?: string): Promise<Catalog> {
 export async function fromStore(store: RDF.Store[]): Promise<Catalog> {
   // Collect all properties for SELECT and GROUP BY so we can flatten the schema:url values into a single value.
   const properties =
-    '?dataset ?name ?description ?creator ?creatorName ?creatorAlternateName ?distribution ?endpointUrl ?searchQuery ?lookupQuery ?reconciliationUrlTemplate ?alternateName';
+    '?dataset ?name ?description ?creator ?creatorName ?creatorAlternateName ?distribution ?endpointUrl ?searchQuery ?lookupQuery ?reconciliationUrlTemplate ?alternateName ?inLanguage';
   const query = `
       PREFIX schema: <http://schema.org/>
         SELECT ${properties} (GROUP_CONCAT(?url) as ?url)  WHERE {
           ?dataset a schema:Dataset ;
             schema:name ?name ;
             schema:description ?description ;
+            schema:inLanguage ?inLanguage ;
             schema:creator ?creator ;
             schema:distribution ?distribution ;
             schema:url ?url .
-          OPTIONAL { ?dataset schema:alternateName ?alternateName . }
           ?creator schema:name ?creatorName ;
             schema:alternateName ?creatorAlternateName .
           ?distribution schema:encodingFormat "application/sparql-query" ;
@@ -52,7 +52,8 @@ export async function fromStore(store: RDF.Store[]): Promise<Catalog> {
                 ?distribution schema:potentialAction/schema:target ?entryPoint .
                 ?entryPoint schema:actionApplication ?reconciliationIri ;
                     schema:urlTemplate ?reconciliationUrlTemplate .
-            } 
+            }
+          OPTIONAL { ?dataset schema:alternateName ?alternateName . }
         }
         GROUP BY ${properties}
         ORDER BY LCASE(?name)`;
@@ -109,7 +110,8 @@ export async function fromStore(store: RDF.Store[]): Promise<Catalog> {
               ]
             ),
           ],
-          bindings.get('alternateName')?.value
+          bindings.get('alternateName')?.value,
+          bindings.get('inLanguage')?.value
         )
       );
     });

diff --git a/packages/network-of-terms-catalog/test/catalog.test.ts b/packages/network-of-terms-catalog/test/catalog.test.ts
@@ -42,6 +42,7 @@ describe('Catalog', () => {
       new IRI('https://data.cultureelerfgoed.nl/term/id/cht/'),
     ]);
     expect(cht.alternateName).toEqual('CHT');
+    expect(cht.inLanguage).toEqual('nl');
     expect(cht.creators[0].name).toEqual(
       'Rijksdienst voor het Cultureel Erfgoed'
     );