WIP inner join 2

halestudio · Feb 7, 2024 · 9fb0cc4 · 9fb0cc4
1 parent c6e943c
commit 9fb0cc4
Show file tree

Hide file tree

Showing 30 changed files with 867 additions and 12 deletions.
diff --git a/common/plugins/eu.esdihumboldt.hale.common.align/plugin.xml b/common/plugins/eu.esdihumboldt.hale.common.align/plugin.xml
@@ -375,6 +375,24 @@
             <parameterBinding>
             </parameterBinding>
          </functionParameter>
+         <functionParameter
+               deprecated="false"
+               description="If enabled a result is created only if all Join conditions are met"
+               label="Inner Join"
+               maxOccurrence="1"
+               minOccurrence="0"
+               name="innerJoin"
+               scriptable="false">
+            <parameterBinding
+                  class="java.lang.Boolean">
+            </parameterBinding>
+            <valueDescriptor
+                  default="false"
+                  defaultDescription="Default behavior is that a result is created for every instance of the Join&apos;s focus type"
+                  sample="true"
+                  sampleDescription="Only create a result from the Join if there are linked instances for all Join conditions">
+            </valueDescriptor>
+         </functionParameter>
       </typeFunction>
       <propertyFunction
             category="eu.esdihumboldt.hale.align.general"

diff --git a/...n/src/eu/esdihumboldt/hale/common/align/model/functions/explanations/JoinExplanation.java b/...n/src/eu/esdihumboldt/hale/common/align/model/functions/explanations/JoinExplanation.java
@@ -72,6 +72,19 @@ protected String getExplanation(Cell cell, boolean html, ServiceProvider service
 			}
 			sb.append('\n');
 
+			// inner join
+
+			boolean innerJoin = CellUtil.getFirstParameter(cell, PARAMETER_INNER_JOIN)
+					.as(Boolean.class, false);
+			if (innerJoin) {
+				sb.append(getMessage("innerjoin", locale));
+			}
+			else {
+				sb.append(MessageFormat.format(getMessage("leftjoin", locale),
+						formatEntity(join.getTypes().get(0), html, false, locale)));
+			}
+			sb.append("\n\n");
+
 			// finalize
 
 			String explanation = sb.toString();

diff --git a/...eu/esdihumboldt/hale/common/align/model/functions/explanations/JoinExplanation.properties b/...eu/esdihumboldt/hale/common/align/model/functions/explanations/JoinExplanation.properties
@@ -1 +1,3 @@
 main=Joins the types {0} based on the following conditions:
+innerjoin=The transformation is only done for instances where all of the conditions are met.
+leftjoin=The transformation is done for every instance of type {0}, regardless which join conditions are met.
diff --git a/...esdihumboldt/hale/common/align/model/functions/explanations/JoinExplanation_de.properties b/...esdihumboldt/hale/common/align/model/functions/explanations/JoinExplanation_de.properties
@@ -1 +1,3 @@
 main=Führt Objekte der Objektarten {0} unter folgenden Bedingungen zusammen:
+innerjoin=Die Transformation wird nur für Objekte ausgeführt, für die alle Bedingungen erfüllt sind.
+leftjoin=Die Transformation wird für alle Objekte vom Typ {0} ausgeführt, unabhängig davon welche Bedingungen erfüllt werden.
diff --git a/...oldt.cst.functions.core/src/eu/esdihumboldt/cst/functions/core/join/IndexJoinHandler.java b/...oldt.cst.functions.core/src/eu/esdihumboldt/cst/functions/core/join/IndexJoinHandler.java
@@ -67,7 +67,7 @@ public ResourceIterator<FamilyInstance> partitionInstances(InstanceCollection in
 			String transformationIdentifier, TransformationEngine engine,
 			ListMultimap<String, ParameterValue> transformationParameters,
 			Map<String, String> executionParameters, TransformationLog log)
-					throws TransformationException {
+			throws TransformationException {
 
 		if (transformationParameters == null
 				|| !transformationParameters.containsKey(PARAMETER_JOIN)
@@ -122,7 +122,13 @@ public ResourceIterator<FamilyInstance> partitionInstances(InstanceCollection in
 			}
 		}
 
-		return new IndexJoinIterator(startInstances, joinDefinition, indexService);
+		boolean innerJoin = false; // default to false if not specified
+		List<ParameterValue> innerJoinValues = transformationParameters.get(PARAMETER_INNER_JOIN);
+		if (!innerJoinValues.isEmpty()) {
+			innerJoin = innerJoinValues.get(0).as(Boolean.class, innerJoin);
+		}
+
+		return new IndexJoinIterator(startInstances, joinDefinition, indexService, innerJoin);
 	}
 
 	/**

diff --git a/...ldt.cst.functions.core/src/eu/esdihumboldt/cst/functions/core/join/IndexJoinIterator.java b/...ldt.cst.functions.core/src/eu/esdihumboldt/cst/functions/core/join/IndexJoinIterator.java
@@ -36,46 +36,51 @@
 import eu.esdihumboldt.hale.common.instance.index.InstanceIndexService;
 import eu.esdihumboldt.hale.common.instance.model.FamilyInstance;
 import eu.esdihumboldt.hale.common.instance.model.ResolvableInstanceReference;
-import eu.esdihumboldt.hale.common.instance.model.impl.GenericResourceIteratorAdapter;
+import eu.esdihumboldt.hale.common.instance.model.impl.FilterResourceIteratorAdapter;
 
 /**
  * Iterator used by {@link IndexJoinHandler}
  * 
  * @author Florian Esser
  */
 class IndexJoinIterator
-		extends GenericResourceIteratorAdapter<ResolvableInstanceReference, FamilyInstance> {
+		extends FilterResourceIteratorAdapter<ResolvableInstanceReference, FamilyInstance> {
 
 	private final JoinDefinition joinDefinition;
 	private final InstanceIndexService index;
 
+	private final boolean innerJoin;
+
 	protected IndexJoinIterator(Collection<ResolvableInstanceReference> startInstances,
-			JoinDefinition joinDefinition, InstanceIndexService index) {
+			JoinDefinition joinDefinition, InstanceIndexService index, boolean innerJoin) {
 		super(startInstances.iterator());
 		this.joinDefinition = joinDefinition;
 		this.index = index;
+		this.innerJoin = innerJoin;
 	}
 
-	/**
-	 * @see eu.esdihumboldt.hale.common.instance.model.impl.GenericResourceIteratorAdapter#convert(java.lang.Object)
-	 */
 	@Override
 	protected FamilyInstance convert(ResolvableInstanceReference next) {
 		FamilyInstance base = new FamilyInstanceImpl(next.resolve());
 		base.getId();
 		FamilyInstance[] currentInstances = new FamilyInstance[joinDefinition.directParent.length];
 		currentInstances[0] = base;
 
-		join(currentInstances, 0);
+		if (!join(currentInstances, 0)) {
+			// skip this instance
+			return null;
+		}
 
 		return base;
 	}
 
 	/**
 	 * Joins all direct children of the given type to currentInstances.
+	 * 
+	 * @return if the instance should be skipped
 	 */
 	@SuppressWarnings("javadoc")
-	private void join(FamilyInstance[] currentInstances, int currentType) {
+	private boolean join(FamilyInstance[] currentInstances, int currentType) {
 		// Join all types that are direct children of the last type.
 		for (int i = currentType + 1; i < joinDefinition.directParent.length; i++) {
 			if (joinDefinition.directParent[i] == currentType) {
@@ -148,12 +153,21 @@ private void join(FamilyInstance[] currentInstances, int currentType) {
 						child = new FamilyInstanceImpl(ref.resolve());
 						parent.addChild(child);
 						currentInstances[i] = child;
-						join(currentInstances, i);
+						if (!join(currentInstances, i)) {
+							return false;
+						}
 					}
 					currentInstances[i] = null;
 				}
+				else {
+					if (innerJoin) {
+						return false;
+					}
+				}
 			}
 		}
+
+		return true;
 	}
 
 	/**

diff --git a/cst/plugins/eu.esdihumboldt.cst.functions.groovy/plugin.xml b/cst/plugins/eu.esdihumboldt.cst.functions.groovy/plugin.xml
@@ -276,6 +276,24 @@
                   ref="text">
             </parameterComplexValue>
          </functionParameter>
+         <functionParameter
+               deprecated="false"
+               description="If enabled a result is created only if all Join conditions are met"
+               label="Inner Join"
+               maxOccurrence="1"
+               minOccurrence="0"
+               name="innerJoin"
+               scriptable="false">
+            <parameterBinding
+                  class="java.lang.Boolean">
+            </parameterBinding>
+            <valueDescriptor
+                  default="false"
+                  defaultDescription="Default behavior is that a result is created for every instance of the Join&apos;s focus type"
+                  sample="true"
+                  sampleDescription="Only create a result from the Join if there are linked instances for all Join conditions">
+            </valueDescriptor>
+         </functionParameter>
       </typeFunction>
    </extension>
    <extension

diff --git a/...dihumboldt.cst.test/src/eu/esdihumboldt/cst/internal/ConceptualSchemaTransformerTest.java b/...dihumboldt.cst.test/src/eu/esdihumboldt/cst/internal/ConceptualSchemaTransformerTest.java
@@ -354,6 +354,40 @@ public void testInnerJoin() throws Exception {
 		testTransform(TransformationExamples.getExample(TransformationExamples.INNER_JOIN));
 	}
 
+	/**
+	 * Test a Join that has the innerJoin flag enabled. Multiple types are
+	 * joined to the join focus, not joins at second level.
+	 * 
+	 * @throws Exception if an error occurs executing the test
+	 */
+	@Test
+	public void testInnerJoinFirstLevel() throws Exception {
+		testTransform(
+				TransformationExamples.getExample(TransformationExamples.INNER_JOIN_FIRST_LEVEL));
+	}
+
+	/**
+	 * Test a Join that has the innerJoin flag enabled. Conditions are bound to
+	 * some of the joined types.
+	 * 
+	 * @throws Exception if an error occurs executing the test
+	 */
+	@Test
+	public void testInnerJoinConditions() throws Exception {
+		testTransform(
+				TransformationExamples.getExample(TransformationExamples.INNER_JOIN_CONDITIONS));
+	}
+
+	/**
+	 * Test a Join that does not have the innerJoin flag enabled.
+	 * 
+	 * @throws Exception if an error occurs executing the test
+	 */
+	@Test
+	public void testJoin() throws Exception {
+		testTransform(TransformationExamples.getExample(TransformationExamples.JOIN));
+	}
+
 	@Override
 	protected List<Instance> transformData(TransformationExample example) throws Exception {
 		ConceptualSchemaTransformer transformer = new ConceptualSchemaTransformer();

diff --git a/...ugins/eu.esdihumboldt.cst.test/src/eu/esdihumboldt/cst/test/TransformationExamples.groovy b/...ugins/eu.esdihumboldt.cst.test/src/eu/esdihumboldt/cst/test/TransformationExamples.groovy
@@ -113,6 +113,9 @@ abstract class TransformationExamples {
 	public static final String XSL_XPATH_1 = 'xpath1'
 
 	public static final String INNER_JOIN = 'inner_join'
+	public static final String INNER_JOIN_FIRST_LEVEL = 'inner_join_first_level'
+	public static final String INNER_JOIN_CONDITIONS = 'inner_join_conditions'
+	public static final String JOIN = 'join'
 
 	/**
 	 * Internal example map.
@@ -230,7 +233,10 @@ abstract class TransformationExamples {
 		(XSL_XPATH_1): defaultExample(XSL_XPATH_1),
 
 		// builder based examples
-		(INNER_JOIN): builderExample(INNER_JOIN)
+		(INNER_JOIN): builderExample(INNER_JOIN),
+		(INNER_JOIN_FIRST_LEVEL): builderExample(INNER_JOIN_FIRST_LEVEL),
+		(INNER_JOIN_CONDITIONS): builderExample(INNER_JOIN_CONDITIONS),
+		(JOIN): builderExample(JOIN)
 	];
 
 	static def defaultExample(String folder) {

diff --git a/cst/plugins/eu.esdihumboldt.cst.test/testdata/inner_join_conditions/mapping.halex b/cst/plugins/eu.esdihumboldt.cst.test/testdata/inner_join_conditions/mapping.halex
@@ -0,0 +1,33 @@
+<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
+<hale-project version="5.1.0.qualifier">
+    <name>mapping</name>
+    <author>hale</author>
+    <created>2024-02-07T10:28:58.718+01:00</created>
+    <modified>2024-02-07T13:02:59.850+01:00</modified>
+    <save-config action-id="project.save" provider-id="eu.esdihumboldt.hale.io.project.hale25.xml.writer">
+        <setting name="charset">UTF-8</setting>
+        <setting name="projectFiles.separate">false</setting>
+        <setting name="contentType">eu.esdihumboldt.hale.io.project.hale25.xml</setting>
+        <setting name="target">file:/home/simon/repos/hale/cst/plugins/eu.esdihumboldt.cst.test/testdata/inner_join_conditions/mapping.halex</setting>
+    </save-config>
+    <resource action-id="eu.esdihumboldt.hale.io.schema.read.source" provider-id="eu.esdihumboldt.hale.io.schemabuilder">
+        <setting name="charset">UTF-8</setting>
+        <setting name="resourceId">ec2ec25e-5940-4b91-8ccd-9ea9455a3dbc</setting>
+        <setting name="source">source-schema.groovy</setting>
+        <setting name="contentType">eu.esdihumboldt.hale.io.schemabuilder</setting>
+    </resource>
+    <resource action-id="eu.esdihumboldt.hale.io.schema.read.target" provider-id="eu.esdihumboldt.hale.io.schemabuilder">
+        <setting name="charset">UTF-8</setting>
+        <setting name="resourceId">646e49d6-a2f9-462c-8bfb-e6a2a47abef1</setting>
+        <setting name="source">target-schema.groovy</setting>
+        <setting name="contentType">eu.esdihumboldt.hale.io.schemabuilder</setting>
+    </resource>
+    <resource action-id="eu.esdihumboldt.hale.io.instance.read.source" provider-id="eu.esdihumboldt.hale.io.instancebuilder">
+        <setting name="charset">UTF-8</setting>
+        <setting name="resourceId">95bd5843-dfc0-43be-a901-03e7b843d44a</setting>
+        <setting name="source">source-instances.groovy</setting>
+        <setting name="contentType">eu.esdihumboldt.hale.io.instancebuilder</setting>
+    </resource>
+    <file name="alignment.xml" location="mapping.halex.alignment.xml"/>
+    <file name="styles.sld" location="mapping.halex.styles.sld"/>
+</hale-project>