Merge pull request #14 from viaacode/arc-2448-add-preferences-to-kg-p…

…ostgres-etl Arc 2448 add preferences to kg postgres etl
viaacode · Sep 2, 2024 · d4cc400 · d4cc400
2 parents e2c3200 + d8a3f14
commit d4cc400
Show file tree

Hide file tree

Showing 5 changed files with 59 additions and 38 deletions.
diff --git a/queries/graph.carrier.sparql b/queries/graph.carrier.sparql
@@ -31,11 +31,9 @@ SELECT DISTINCT
 ?created_at
 ?updated_at
 WHERE {
-    ?intellectual_entity_id rel:isr ?rep; dct:format ?format.
-    ?rep premis:storedAt ?id.
-    ?id a haObj:PhysicalCarrier;
-    schema:name ?schema_name;
-    premis:medium/skos:prefLabel ?premis_medium.
+    ?intellectual_entity_id rel:isr ?rep;  
+      schema:license ?schema_license;
+      dct:format ?format.
 
     # Do not include images
     VALUES ?format {
@@ -48,21 +46,25 @@ WHERE {
       "newspaperpage"
       "newspaper"
     } 
+
     # Only include the right licenses
-    FILTER EXISTS {
-      ?intellectual_entity_id schema:license ?schema_license
-      VALUES ?schema_license { 
-        haLicId:VIAA-PUBLIEK-METADATA-LTD 
-        haLicId:VIAA-PUBLIEK-METADATA-ALL
-        haLicId:VIAA-PUBLIEK-CONTENT
-        haLicId:BEZOEKERTOOL-METADATA-ALL
-        haLicId:BEZOEKERTOOL-CONTENT 
-        haLicId:VIAA-INTRA_CP-METADATA-ALL 
-        haLicId:VIAA-INTRA_CP-CONTENT 
-        haLicId:Publiek-Domein
-        haLicId:COPYRIGHT-UNDETERMINED
-    }
+    VALUES ?schema_license { 
+      haLicId:VIAA-PUBLIEK-METADATA-LTD 
+      haLicId:VIAA-PUBLIEK-METADATA-ALL
+      haLicId:VIAA-PUBLIEK-CONTENT
+      haLicId:BEZOEKERTOOL-METADATA-ALL
+      haLicId:BEZOEKERTOOL-CONTENT 
+      haLicId:VIAA-INTRA_CP-METADATA-ALL 
+      haLicId:VIAA-INTRA_CP-CONTENT 
+      haLicId:Publiek-Domein
+      haLicId:COPYRIGHT-UNDETERMINED
     }
+
+    ?rep premis:storedAt ?id.
+
+    ?id a haObj:PhysicalCarrier;
+    schema:name ?schema_name;
+    premis:medium/skos:prefLabel ?premis_medium.
 
     FILTER (langMatches(lang(?schema_name), "nl") && langMatches(lang(?premis_medium), "en"))
 

diff --git a/queries/graph.file.sparql b/queries/graph.file.sparql
@@ -33,10 +33,11 @@ WHERE {
         premis:originalName ?premis_original_name;
         premis:storedAt/rdf:value ?premis_stored_at ;
         ebucore:hasMimeType ?ebucore_has_mime_type ;
-        prov:wasDerivedFrom/schema:dateModified ?modified ;
-        (rel:isi|ebucore:hasMediaFragment)/rel:rep ?ie.
+        prov:wasDerivedFrom/schema:dateModified ?modified .
+        # (rel:isi|ebucore:hasMediaFragment)/rel:rep ?ie.
 
-        ?ie schema:license ?schema_license; dct:format ?format .
+    #     ?ie dct:format ?format .
+    #     ?ie schema:license ?schema_license
 
     # VALUES ?schema_license { 
     #     haLicId:VIAA-PUBLIEK-METADATA-LTD 
@@ -51,16 +52,16 @@ WHERE {
     # }
 
     # Do not include images
-    VALUES ?format {
-      #"image"
-      "video"
-      "videofragment"
-      "audiofragment"
-      "audio"
-      "film"
-      "newspaperpage"
-      "newspaper"
-    } 
+    # VALUES ?format {
+    #   #"image"
+    #   "video"
+    #   "videofragment"
+    #   "audiofragment"
+    #   "audio"
+    #   "film"
+    #   "newspaperpage"
+    #   "newspaper"
+    # } 
 
     FILTER (!isLiteral(?since) || ?modified >= ?since )
 
@@ -79,5 +80,5 @@ WHERE {
       ?id schema:duration ?schema_duration_res.
       BIND (xsd:double(?schema_duration_res) AS ?schema_duration)
     }
-
-}
+}
+ORDER BY ?id 
diff --git a/queries/graph.includes.sparql b/queries/graph.includes.sparql
@@ -63,4 +63,5 @@ WHERE {
 
     ?file_id prov:wasDerivedFrom/schema:dateModified ?modified .
     FILTER (!isLiteral(?since) || ?modified >= ?since )
-}
+}
+ORDER BY ?representation_id ?file_id
diff --git a/queries/graph.organization.sparql b/queries/graph.organization.sparql
@@ -9,8 +9,9 @@ SELECT DISTINCT
 ?id
 ?dcterms_description
 ?foaf_homepage
-?ha_org_allows_bzt
-?ha_org_allows_overlay
+# ?ha_org_allows_bzt
+# ?ha_org_allows_iiif
+# ?ha_org_allows_overlay
 ?ha_org_has_logo
 ?ha_org_request_form
 ?ha_org_sector
@@ -28,8 +29,9 @@ WHERE {
   OPTIONAL { ?id dct:description ?dcterms_description }
   OPTIONAL { ?id foaf:homepage ?foaf_homepage }
   OPTIONAL { ?id haOrg:hasLogo ?ha_org_has_logo }
-  BIND (EXISTS { ?id haOrg:hasPreference haPref:logo-embedding } AS ?ha_org_allows_overlay)
-  BIND (EXISTS { ?id haOrg:hasPreference haPref:visitor-space-publication } AS ?ha_org_allows_bzt)
+  # BIND (EXISTS { ?id haOrg:hasPreference haPref:logo-embedding } AS ?ha_org_allows_overlay)
+  # BIND (EXISTS { ?id haOrg:hasPreference haPref:visitor-space-publication } AS ?ha_org_allows_bzt)
+  # BIND (EXISTS { ?id haOrg:hasPreference haPref:iiif-dissemination } AS ?ha_org_allows_iiif)
   OPTIONAL { ?id haOrg:requestForm ?ha_org_request_form } 
   OPTIONAL { ?id haOrg:sector ?ha_org_sector } 
   OPTIONAL { 

diff --git a/queries/graph.organization_has_preference.sparql b/queries/graph.organization_has_preference.sparql
@@ -0,0 +1,15 @@
+# position: 1
+PREFIX skos: <http://www.w3.org/2004/02/skos/core#>
+PREFIX haOrg: <https://data.hetarchief.be/ns/organization/>
+
+SELECT DISTINCT 
+?organization_id
+?ha_pref
+?created_at
+?updated_at
+WHERE {
+  ?organization_id a haOrg:ContentPartner;
+      haOrg:hasPreference ?ha_pref_res .
+
+   BIND (REPLACE(STR(?ha_pref_res), "^.*/([^/]*)$", "$1") as ?ha_pref)
+}