Refactor SPARQL updates for kingdom assignments and introduce new scripts for propagation

Author: retog

.github/workflows/main.yml

@@ -47,7 +47,17 @@ jobs:
 
           tdb2.tdbloader --loc=./tdbstore col.nt
           rm col.nt
-          tdb2.tdbupdate --loc=./tdbstore --update=add-kingdoms.sparql
+
+          echo "Running add-kingdoms-root.sparql"
+          tdb2.tdbupdate --loc=./tdbstore --update=add-kingdoms-root.sparql
+
+          for i in {1..6}; do
+            echo "Running propagate-kingdoms.sparql (iteration $i)"
+            tdb2.tdbupdate --loc=./tdbstore --update=propagate-kingdoms.sparql
+          done
+
+          echo "Running propagate-kingdoms-acceptedname.sparql"
+          tdb2.tdbupdate --loc=./tdbstore --update=propagate-kingdoms-acceptedname.sparql
 
           tdb2.tdbdump --loc=./tdbstore | gzip > col.nt.gz
 

add-kingdoms-root.sparql

@@ -0,0 +1,11 @@
+PREFIX dwc: <http://rs.tdwg.org/dwc/terms/>
+
+INSERT {
+  ?taxon dwc:kingdom ?kingdom .
+} WHERE {
+  ?taxon dwc:taxonRank "kingdom" ;
+         dwc:scientificName ?full_name .
+  OPTIONAL { ?taxon dwc:scientificNameAuthorship ?authority . }
+  BIND(REPLACE(?full_name, CONCAT(" ", COALESCE(?authority, ""), "$"), "") AS ?kingdom)
+  # In some cases (8 of 23), the kingdom has an authority. the above should trim it.
+};

add-kingdoms.sparql

@@ -0,0 +1,10 @@
+PREFIX dwc: <http://rs.tdwg.org/dwc/terms/>
+
+INSERT {
+  ?taxon dwc:kingdom ?kingdom .
+} WHERE {
+  ?taxon dwc:acceptedName ?an .
+  ?an dwc:kingdom ?kingdom .
+  FILTER NOT EXISTS { ?taxon dwc:kingdom ?any }
+};
+

propagate-kingdoms-acceptedname.sparql

@@ -0,0 +1,10 @@
+PREFIX dwc: <http://rs.tdwg.org/dwc/terms/>
+
+INSERT {
+  ?taxon dwc:kingdom ?kingdom .
+} WHERE {
+  ?taxon dwc:parent ?parent .
+  ?parent dwc:kingdom ?kingdom .
+  FILTER NOT EXISTS { ?taxon dwc:kingdom ?any }
+};
+