From f35451ce22efca8927a1551db559004dae8c1158 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Kalle=20W=C3=A5hlin?=
 <72360110+kwahlin@users.noreply.github.com>
Date: Wed, 20 Nov 2024 17:49:12 +0100
Subject: [PATCH] Safe guard against matching non-existing subtypes (#1525)

---
 .../whelk/datatool/form/MatchForm.groovy       | 12 +++++-------
 .../whelk/datatool/form/MatchFormSpec.groovy   | 18 ++++++++++++++++++
 2 files changed, 23 insertions(+), 7 deletions(-)

diff --git a/whelktool/src/main/groovy/whelk/datatool/form/MatchForm.groovy b/whelktool/src/main/groovy/whelk/datatool/form/MatchForm.groovy
index 0e828ba6a9..85688e7905 100644
--- a/whelktool/src/main/groovy/whelk/datatool/form/MatchForm.groovy
+++ b/whelktool/src/main/groovy/whelk/datatool/form/MatchForm.groovy
@@ -1,6 +1,5 @@
 package whelk.datatool.form
 
-import groovy.transform.Memoized
 import groovy.transform.PackageScope
 import whelk.Document
 import whelk.JsonLd
@@ -9,7 +8,6 @@ import whelk.datatool.util.DocumentComparator
 import whelk.datatool.util.IdLoader
 import whelk.util.DocumentUtil
 
-import static whelk.JsonLd.GRAPH_KEY
 import static whelk.JsonLd.ID_KEY
 import static whelk.JsonLd.RECORD_KEY
 import static whelk.JsonLd.RECORD_TYPE
@@ -78,9 +76,7 @@ class MatchForm {
         thing[ID_KEY] = getThingTmpId()
         record[THING_KEY] = [(ID_KEY): getThingTmpId()]
 
-        Map graph = [(GRAPH_KEY): [record, thing]]
-
-        String ttl = toTurtleNoPrelude(graph, context)
+        String ttl = toTurtleNoPrelude([record, thing], context)
 
         return insertTypeMappings(insertIdMappings(insertVars(ttl)))
     }
@@ -164,7 +160,7 @@ class MatchForm {
                 if (node[TYPE_KEY] == ANY_TYPE) {
                     node.remove(TYPE_KEY)
                 }
-                if (asList(node.remove(MATCHING_MODE)).contains(SUBTYPES)) {
+                if (asList(node.remove(MATCHING_MODE)).contains(SUBTYPES) && baseTypeToSubtypes.containsKey(node[TYPE_KEY])) {
                     def baseType = node.remove(TYPE_KEY)
                     node[HAS_BASE_TYPE_TMP] = baseType
                 }
@@ -300,7 +296,9 @@ class MatchForm {
             if (node instanceof Map && node.containsKey(MATCHING_MODE) && ((List) node[MATCHING_MODE]).contains(SUBTYPES)) {
                 def baseType = (String) node[TYPE_KEY]
                 Set<String> subTypes = getSubtypes(baseType, jsonLd) as Set
-                mappings[baseType] = subTypes
+                if (!subTypes.isEmpty()) {
+                    mappings[baseType] = subTypes
+                }
                 return new DocumentUtil.Nop()
             }
         }
diff --git a/whelktool/src/test/groovy/whelk/datatool/form/MatchFormSpec.groovy b/whelktool/src/test/groovy/whelk/datatool/form/MatchFormSpec.groovy
index 7cfa53db6d..9a8612a4a9 100644
--- a/whelktool/src/test/groovy/whelk/datatool/form/MatchFormSpec.groovy
+++ b/whelktool/src/test/groovy/whelk/datatool/form/MatchFormSpec.groovy
@@ -175,4 +175,22 @@ class MatchFormSpec extends Specification {
         expect:
         transform.getSparqlPattern(context) == expectedPattern
     }
+
+    def "form to sparql pattern: no subtypes to match"() {
+        given:
+        def form = [
+                'bulk:formBlankNodeId': '#1',
+                '@type'               : 'T1',
+                'bulk:matchingMode'   : ['bulk:Subtypes']
+        ]
+
+        def expectedPattern = "?graph :mainEntity ?1 .\n" +
+                "\n" +
+                "?1 a :T1 ."
+
+        def transform = new MatchForm(form)
+
+        expect:
+        transform.getSparqlPattern(context) == expectedPattern
+    }
 }