[New] Make sure that only transformed datasets are deduplicated

I.e. do not touch dataset that does not start with "http://onto.fel.cvut.cz/dataset--20"
kbss-cvut · Oct 10, 2023 · 3e4ee07 · 3e4ee07
1 parent 281386a
commit 3e4ee07
Show file tree

Hide file tree

Showing 2 changed files with 2 additions and 1 deletion.
diff --git a/aircraft-maintenance-planning-model/data/bin/4-graphdb-deduplicate-repository.sh b/aircraft-maintenance-planning-model/data/bin/4-graphdb-deduplicate-repository.sh
@@ -12,7 +12,7 @@ NETRC_FILE=../private/netrc
 SERVER_URL="https://graphdb.onto.fel.cvut.cz"
 GRAPHDB_REPOSITORY_URL="${SERVER_URL}/repositories/${REPOSITORY_ID}"
 
-curl --netrc-file "$NETRC_FILE" $GRAPHDB_REPOSITORY_URL/contexts 2>/dev/null | tail +2 | sed -e "s/\r//g" | sort -r > private/graphs.csv
+curl --netrc-file "$NETRC_FILE" $GRAPHDB_REPOSITORY_URL/contexts 2>/dev/null | grep dataset--20 | tail +2 | sed -e "s/\r//g" | sort -r > private/graphs.csv
 
 cat private/graphs.csv | while read GRAPH; do
 	rm ./queries/deduplicate-triples.uq

diff --git a/aircraft-maintenance-planning-model/data/queries/deduplicate-triples-simple.uq b/aircraft-maintenance-planning-model/data/queries/deduplicate-triples-simple.uq
@@ -11,6 +11,7 @@ DELETE {
             GRAPH ?olderG {
                 ?s ?p ?o .
             }
+            FILTER(strstarts(str(?olderG),"http://onto.fel.cvut.cz/dataset--20"))
             FILTER(str(?olderG) < str(?newerG))
         } LIMIT 100000
     }