feat(geopackage): prefer non-GML namespace on multiple candidates for…

… property ING-4543
halestudio · Nov 29, 2024 · 9c4ecbe · 9c4ecbe
1 parent 0089368
commit 9c4ecbe
Show file tree

Hide file tree

Showing 4 changed files with 108 additions and 0 deletions.
diff --git a/io/plugins/eu.esdihumboldt.hale.io.geopackage/resources/test/testdata/gml-conflict.gpkg b/io/plugins/eu.esdihumboldt.hale.io.geopackage/resources/test/testdata/gml-conflict.gpkg
diff --git a/io/plugins/eu.esdihumboldt.hale.io.geopackage/resources/test/testdata/gml-conflict.xsd b/io/plugins/eu.esdihumboldt.hale.io.geopackage/resources/test/testdata/gml-conflict.xsd
@@ -0,0 +1,33 @@
+<?xml version="1.0" encoding="UTF-8" standalone="no"?>
+<schema xmlns="http://www.w3.org/2001/XMLSchema" xmlns:gml="http://www.opengis.net/gml/3.2" xmlns:planAtlas-="https://www.geoportal-raumordnung-bw.de/planAtlas-rp" elementFormDefault="qualified" targetNamespace="https://www.geoportal-raumordnung-bw.de/planAtlas-rp">
+    <import namespace="http://www.opengis.net/gml/3.2" schemaLocation="http://schemas.opengis.net/gml/3.2.1/gml.xsd"/>
+    <element name="ING-4543" substitutionGroup="gml:AbstractFeature" type="planAtlas-:ING-4543"/>
+    <complexType name="ING-4543">
+        <complexContent>
+            <extension base="gml:AbstractFeatureType">
+                <sequence>
+                    <element maxOccurs="1" minOccurs="0" name="bereich" nillable="false" type="integer"/>
+                    <element maxOccurs="1" minOccurs="0" name="oa_nr" nillable="false" type="integer"/>
+                    <element maxOccurs="1" minOccurs="0" name="oa_nr_t" nillable="false" type="string"/>
+                    <element maxOccurs="1" minOccurs="0" name="klasse" nillable="false" type="integer"/>
+                    <element maxOccurs="1" minOccurs="0" name="klasse_t" nillable="false" type="string"/>
+                    <element maxOccurs="1" minOccurs="0" name="bindung" nillable="false" type="integer"/>
+                    <element maxOccurs="1" minOccurs="0" name="bindung_t" nillable="false" type="string"/>
+                    <element maxOccurs="1" minOccurs="0" name="status" nillable="false" type="integer"/>
+                    <element maxOccurs="1" minOccurs="0" name="status_t" nillable="false" type="string"/>
+                    <element maxOccurs="1" minOccurs="0" name="kategorie" nillable="false" type="integer"/>
+                    <element maxOccurs="1" minOccurs="0" name="kategori_t" nillable="false" type="string"/>
+                    <element maxOccurs="1" minOccurs="0" name="erst_id" nillable="false" type="integer"/>
+                    <element maxOccurs="1" minOccurs="0" name="erst_id_t" nillable="false" type="string"/>
+                    <element maxOccurs="1" minOccurs="0" name="name" nillable="false" type="string"/>
+                    <element maxOccurs="1" minOccurs="0" name="bemerkung" nillable="false" type="string"/>
+                    <element maxOccurs="1" minOccurs="0" name="inkraft" nillable="false" type="date"/>
+                    <element maxOccurs="1" minOccurs="0" name="massstab" nillable="false" type="integer"/>
+                    <element maxOccurs="1" minOccurs="0" name="nr" nillable="false" type="integer"/>
+                    <element maxOccurs="1" minOccurs="0" name="id" nillable="false" type="string"/>
+                    <element maxOccurs="1" minOccurs="0" name="geometrie" nillable="false" type="gml:MultiCurvePropertyType"/>
+                </sequence>
+            </extension>
+        </complexContent>
+    </complexType>
+</schema>
diff --git a/...io.geopackage/src/eu/esdihumboldt/hale/io/geopackage/internal/TableInstanceBuilder.groovy b/...io.geopackage/src/eu/esdihumboldt/hale/io/geopackage/internal/TableInstanceBuilder.groovy
@@ -12,6 +12,8 @@
  */
 package eu.esdihumboldt.hale.io.geopackage.internal
 
+import static eu.esdihumboldt.hale.common.schema.SchemaConstants.GML_NAMESPACE_CORE
+
 import groovy.transform.CompileStatic
 
 import java.sql.SQLException
@@ -102,6 +104,16 @@ class TableInstanceBuilder {
 				cand = candidates.find { it.getConstraint(Cardinality).getMinOccurs() > 0 && !it.getConstraint(NillableFlag).isEnabled() }
 			}
 
+			if (cand == null && candidates.size() == 2) {
+				// prefer property that is not in GML namespace, if the other is in a GML namespace
+				if (candidates[0].name.namespaceURI.startsWith(GML_NAMESPACE_CORE)) {
+					cand = candidates[1]
+				}
+				else if (candidates[1].name.namespaceURI.startsWith(GML_NAMESPACE_CORE)) {
+					cand = candidates[0]
+				}
+			}
+
 			if (cand == null) {
 				// just pick the first one
 				cand = candidates[0]

diff --git a/...io.geopackage/test/eu/esdihumboldt/hale/io/geopackage/GeopackageInstanceReaderTest.groovy b/...io.geopackage/test/eu/esdihumboldt/hale/io/geopackage/GeopackageInstanceReaderTest.groovy
@@ -22,6 +22,8 @@ import groovy.transform.TypeCheckingMode
 
 import java.util.function.Consumer
 
+import javax.xml.namespace.QName
+
 import org.junit.Test
 import org.locationtech.jts.geom.Geometry
 
@@ -36,6 +38,7 @@ import eu.esdihumboldt.hale.common.schema.geometry.GeometryProperty
 import eu.esdihumboldt.hale.common.schema.model.Schema
 import eu.esdihumboldt.hale.common.test.TestUtil
 import eu.esdihumboldt.util.test.AbstractPlatformTest
+import io.qameta.allure.Link
 
 /**
  * Tests loading GeoPackage instances.
@@ -261,4 +264,64 @@ class GeopackageInstanceReaderTest extends AbstractPlatformTest {
 							.isNotNull()
 				} as Consumer)
 	}
+
+	/**
+	 * Test reading Geopackage data based on an XML schema (Use case: Rewrite Geopackage as GML).
+	 */
+	@CompileStatic(TypeCheckingMode.SKIP)
+	@Link(value = "ING-4543", type = "JIRA")
+	@Test
+	public void testReadInstancesXsdGmlConflict() {
+		GeopackageApiTest.withTestFile("testdata/gml-conflict.gpkg") { File file ->
+			Schema xmlSchema = TestUtil.loadSchema(getClass().getClassLoader().getResource("testdata/gml-conflict.xsd").toURI())
+
+			InstanceCollection instances = loadInstances(xmlSchema, file)
+
+			assertNotNull(instances)
+			assertTrue(instances.hasSize())
+
+			// test count
+			assertEquals(3, instances.size())
+
+			// check instances
+			def list = instances.toList()
+			def all =  { new InstanceAccessor(list) }
+
+			def expectedNames = [
+				'Allgäubahn',
+				'Bodensee-Gürtelbahn',
+				'Donaubahn'
+			].toArray()
+
+			// verify that names are present
+			assertThat(all().name.values())
+					.containsExactlyInAnyOrder(expectedNames)
+
+			// verify that correct name property is used
+			def nameProperty = new QName('https://www.geoportal-raumordnung-bw.de/planAtlas-rp', 'name')
+			def names = list.collect { it.getProperty(nameProperty)[0] }.toList()
+			assertThat(names)
+					.containsExactlyInAnyOrder(expectedNames)
+
+			// also check for id element (which is present in the schema in addition to the GML id attribute)
+			def idProperty = new QName('https://www.geoportal-raumordnung-bw.de/planAtlas-rp', 'id')
+			def ids = list.collect { it.getProperty(idProperty)[0] }.toList()
+			assertThat(ids)
+					.containsExactlyInAnyOrder('ln_8164_843_1', 'ln_8164_843_2', 'ln_8164_843_3')
+
+			// GML id attribute should be empty
+			def gmlIdProperty = new QName('http://www.opengis.net/gml/3.2', 'id')
+			def gmlIds = list.collect {
+				def prop = it.getProperty(gmlIdProperty)
+				if (prop) {
+					prop[0]
+				}
+				else {
+					null
+				}
+			}.findAll()
+			assertThat(gmlIds)
+					.isEmpty()
+		}
+	}
 }