Search in sources :

Example 1 with SurvivorshipKeyDefinition

use of org.talend.dataquality.rules.SurvivorshipKeyDefinition in project tdq-studio-se by Talend.

the class AnalysisRecordGroupingUtilsTest method createKeyDefinition.

/**
 * Create KeyDefinition
 *
 * @return
 */
private SurvivorshipKeyDefinition createKeyDefinition(String name, AlgorithmDefinition algorDef) {
    SurvivorshipKeyDefinition createSurvivorshipKeyDefinition = RulesFactory.eINSTANCE.createSurvivorshipKeyDefinition();
    createSurvivorshipKeyDefinition.setName(name);
    createSurvivorshipKeyDefinition.setFunction(algorDef);
    return createSurvivorshipKeyDefinition;
}
Also used : SurvivorshipKeyDefinition(org.talend.dataquality.rules.SurvivorshipKeyDefinition)

Example 2 with SurvivorshipKeyDefinition

use of org.talend.dataquality.rules.SurvivorshipKeyDefinition in project tdq-studio-se by Talend.

the class AnalysisRecordGroupingUtilsTest method testCreateSurvivorShipAlgorithmParams.

/**
 * Test method for
 * {@link org.talend.dq.analysis.AnalysisRecordGroupingUtils#createSurvivorShipAlgorithmParams(org.talend.dataquality.record.linkage.grouping.AnalysisMatchRecordGrouping, org.talend.dataquality.indicators.columnset.RecordMatchingIndicator, Map)
 * .
 */
@Test
public void testCreateSurvivorShipAlgorithmParams() {
    // Survivorshipkey
    RecordMatchingIndicator recordMatchingIndicator = ColumnsetFactory.eINSTANCE.createRecordMatchingIndicator();
    MatchRuleDefinition createMatchRuleDefinition = RulesFactory.eINSTANCE.createMatchRuleDefinition();
    recordMatchingIndicator.setBuiltInMatchRuleDefinition(createMatchRuleDefinition);
    EList<SurvivorshipKeyDefinition> survivorshipKeys = createMatchRuleDefinition.getSurvivorshipKeys();
    AlgorithmDefinition createAlgorithmDefinition = RulesFactory.eINSTANCE.createAlgorithmDefinition();
    // $NON-NLS-1$
    createAlgorithmDefinition.setAlgorithmParameters("");
    // $NON-NLS-1$
    createAlgorithmDefinition.setAlgorithmType("Longest");
    // $NON-NLS-1$
    survivorshipKeys.add(createKeyDefinition("a1", createAlgorithmDefinition));
    createAlgorithmDefinition = RulesFactory.eINSTANCE.createAlgorithmDefinition();
    // $NON-NLS-1$
    createAlgorithmDefinition.setAlgorithmParameters("");
    // $NON-NLS-1$
    createAlgorithmDefinition.setAlgorithmType("Longest");
    // $NON-NLS-1$
    survivorshipKeys.add(createKeyDefinition("a2", createAlgorithmDefinition));
    // DefaultSurvivorship
    EList<DefaultSurvivorshipDefinition> defaultSurvivorshipDefinitions = createMatchRuleDefinition.getDefaultSurvivorshipDefinitions();
    createAlgorithmDefinition = RulesFactory.eINSTANCE.createAlgorithmDefinition();
    // $NON-NLS-1$
    createAlgorithmDefinition.setAlgorithmParameters("");
    // $NON-NLS-1$
    createAlgorithmDefinition.setAlgorithmType("MostCommon");
    // $NON-NLS-1$
    defaultSurvivorshipDefinitions.add(createDefaultsurvivShip("String", createAlgorithmDefinition));
    // DefaultSurvivorship
    EList<ParticularDefaultSurvivorshipDefinitions> particularDefaultSurvivorshipDefinitions = createMatchRuleDefinition.getParticularDefaultSurvivorshipDefinitions();
    createAlgorithmDefinition = RulesFactory.eINSTANCE.createAlgorithmDefinition();
    // $NON-NLS-1$
    createAlgorithmDefinition.setAlgorithmParameters("");
    // $NON-NLS-1$
    createAlgorithmDefinition.setAlgorithmType("Concatenate");
    particularDefaultSurvivorshipDefinitions.add(createParticularDefaultSurvivorshipDefinitions("a2", // $NON-NLS-1$
    createAlgorithmDefinition));
    // init columnMap
    Map<MetadataColumn, String> columnMap = new HashMap<MetadataColumn, String>();
    MetadataColumn col0 = ConnectionFactory.eINSTANCE.createMetadataColumn();
    // $NON-NLS-1$
    col0.setName("a1");
    // $NON-NLS-1$
    col0.setTalendType("id_String");
    // $NON-NLS-1$
    columnMap.put(col0, "0");
    MetadataColumn col1 = ConnectionFactory.eINSTANCE.createMetadataColumn();
    // $NON-NLS-1$
    col1.setTalendType("id_String");
    // $NON-NLS-1$
    col1.setName("a2");
    // $NON-NLS-1$
    columnMap.put(col1, "1");
    MetadataColumn col2 = ConnectionFactory.eINSTANCE.createMetadataColumn();
    // $NON-NLS-1$
    col2.setTalendType("id_String");
    // $NON-NLS-1$
    col2.setName("a3");
    // $NON-NLS-1$
    columnMap.put(col2, "2");
    MatchGroupResultConsumer matchGroupResultConsumer = new MatchGroupResultConsumer(true) {

        @Override
        public void handle(Object row) {
        // no need to implement
        }
    };
    AnalysisMatchRecordGrouping analysisMatchRecordGrouping = new AnalysisMatchRecordGrouping(matchGroupResultConsumer);
    CombinedRecordMatcher combinedRecordMatcher = analysisMatchRecordGrouping.getCombinedRecordMatcher();
    DQMFBRecordMatcher dqmfbRecordMatcher = new DQMFBRecordMatcher(0.9);
    combinedRecordMatcher.getMatchers().add(dqmfbRecordMatcher);
    List<List<Map<String, String>>> multiMatchRules = analysisMatchRecordGrouping.getMultiMatchRules();
    List<Map<String, String>> matchRuleList = new ArrayList<Map<String, String>>();
    Map<String, String> matchKeyMap1 = new HashMap<String, String>();
    Map<String, String> matchKeyMap2 = new HashMap<String, String>();
    Map<String, String> matchKeyMap3 = new HashMap<String, String>();
    matchKeyMap1.put(IRecordGrouping.MATCHING_TYPE, AttributeMatcherType.EXACT.name());
    // change by
    matchKeyMap2.put(IRecordGrouping.MATCHING_TYPE, AttributeMatcherType.DUMMY.name());
    // DefaultSurvivorshipDefinitions
    // change by
    matchKeyMap3.put(IRecordGrouping.MATCHING_TYPE, AttributeMatcherType.DUMMY.name());
    // ParticularDefaultSurvivorshipDefinitions
    matchRuleList.add(matchKeyMap1);
    matchRuleList.add(matchKeyMap2);
    matchRuleList.add(matchKeyMap3);
    // $NON-NLS-1$
    matchKeyMap1.put(IRecordGrouping.MATCH_KEY_NAME, "a1");
    // $NON-NLS-1$
    matchKeyMap2.put(IRecordGrouping.MATCH_KEY_NAME, "a2");
    // $NON-NLS-1$
    matchKeyMap3.put(IRecordGrouping.MATCH_KEY_NAME, "a3");
    // $NON-NLS-1$
    matchKeyMap1.put(IRecordGrouping.COLUMN_IDX, "0");
    // $NON-NLS-1$
    matchKeyMap2.put(IRecordGrouping.COLUMN_IDX, "1");
    // $NON-NLS-1$
    matchKeyMap3.put(IRecordGrouping.COLUMN_IDX, "2");
    multiMatchRules.add(matchRuleList);
    SurvivorShipAlgorithmParams createSurvivorShipAlgorithmParams = AnalysisRecordGroupingUtils.createSurvivorShipAlgorithmParams(analysisMatchRecordGrouping, recordMatchingIndicator, columnMap);
    // $NON-NLS-1$
    Assert.assertEquals(// $NON-NLS-1$
    "The size of SurvivorShipAlgos should be 2", // $NON-NLS-1$
    2, createSurvivorShipAlgorithmParams.getSurviorShipAlgos().length);
    // the size of default survivorshipRules is come from by (column size * default item size)
    Assert.assertEquals("The size of DefaultSurviorshipRules should be 3", 3, // $NON-NLS-1$
    createSurvivorShipAlgorithmParams.getDefaultSurviorshipRules().size());
    Map<IRecordMatcher, SurvivorshipFunction[]> survivorshipAlgosMap = createSurvivorShipAlgorithmParams.getSurvivorshipAlgosMap();
    // $NON-NLS-1$
    Assert.assertEquals("The size of survivorshipAlgosMap should be 1", 1, survivorshipAlgosMap.size());
    SurvivorshipFunction[] survivorshipFunctions = survivorshipAlgosMap.get(dqmfbRecordMatcher);
    // $NON-NLS-1$
    Assert.assertEquals("The size of survivorshipFunctions should be 3", 3, survivorshipFunctions.length);
    // $NON-NLS-1$
    Assert.assertEquals(// $NON-NLS-1$
    "The Algorithm of a1 function should be LONGEST", // $NON-NLS-1$
    SurvivorShipAlgorithmEnum.LONGEST, survivorshipFunctions[0].getSurvivorShipAlgoEnum());
    // $NON-NLS-1$
    Assert.assertEquals(// $NON-NLS-1$
    "The Algorithm of a2 function should be Concatenate", // $NON-NLS-1$
    SurvivorShipAlgorithmEnum.CONCATENATE, survivorshipFunctions[1].getSurvivorShipAlgoEnum());
    // $NON-NLS-1$
    Assert.assertEquals(// $NON-NLS-1$
    "The Algorithm of a3 function should be MostCommon", // $NON-NLS-1$
    SurvivorShipAlgorithmEnum.MOST_COMMON, survivorshipFunctions[2].getSurvivorShipAlgoEnum());
}
Also used : AnalysisMatchRecordGrouping(org.talend.dataquality.record.linkage.grouping.AnalysisMatchRecordGrouping) IRecordMatcher(org.talend.dataquality.record.linkage.record.IRecordMatcher) HashMap(java.util.HashMap) DefaultSurvivorshipDefinition(org.talend.dataquality.rules.DefaultSurvivorshipDefinition) SurvivorShipAlgorithmParams(org.talend.dataquality.record.linkage.grouping.swoosh.SurvivorShipAlgorithmParams) ArrayList(java.util.ArrayList) MatchRuleDefinition(org.talend.dataquality.rules.MatchRuleDefinition) DQMFBRecordMatcher(org.talend.dataquality.record.linkage.grouping.swoosh.DQMFBRecordMatcher) SurvivorshipFunction(org.talend.dataquality.record.linkage.grouping.swoosh.SurvivorShipAlgorithmParams.SurvivorshipFunction) MetadataColumn(org.talend.core.model.metadata.builder.connection.MetadataColumn) AlgorithmDefinition(org.talend.dataquality.rules.AlgorithmDefinition) ArrayList(java.util.ArrayList) EList(org.eclipse.emf.common.util.EList) List(java.util.List) ParticularDefaultSurvivorshipDefinitions(org.talend.dataquality.rules.ParticularDefaultSurvivorshipDefinitions) SurvivorshipKeyDefinition(org.talend.dataquality.rules.SurvivorshipKeyDefinition) RecordMatchingIndicator(org.talend.dataquality.indicators.columnset.RecordMatchingIndicator) MatchGroupResultConsumer(org.talend.dataquality.record.linkage.grouping.MatchGroupResultConsumer) HashMap(java.util.HashMap) Map(java.util.Map) CombinedRecordMatcher(org.talend.dataquality.record.linkage.record.CombinedRecordMatcher) Test(org.junit.Test)

Example 3 with SurvivorshipKeyDefinition

use of org.talend.dataquality.rules.SurvivorshipKeyDefinition in project tdq-studio-se by Talend.

the class AnalysisMatchParameterAdapter method getAllSurvivorshipFunctions.

/*
     * (non-Javadoc)
     * 
     * @see org.talend.dataquality.record.linkage.grouping.adapter.MatchParameterAdapter#getAllSurvivorshipFunctions()
     */
@Override
public List<SurvivorshipFunction> getAllSurvivorshipFunctions() {
    List<SurvivorshipFunction> survFunctions = new ArrayList<SurvivorshipFunction>();
    // Survivorship functions.
    List<SurvivorshipKeyDefinition> survivorshipKeyDefs = recordMatchingIndicator.getBuiltInMatchRuleDefinition().getSurvivorshipKeys();
    for (SurvivorshipKeyDefinition survDef : survivorshipKeyDefs) {
        SurvivorshipFunction func = new SurvivorShipAlgorithmParams().new SurvivorshipFunction();
        func.setSurvivorShipKey(survDef.getName());
        func.setParameter(survDef.getFunction().getAlgorithmParameters());
        func.setSurvivorShipAlgoEnum(SurvivorShipAlgorithmEnum.getTypeBySavedValue(survDef.getFunction().getAlgorithmType()));
        survFunctions.add(func);
    }
    return survFunctions;
}
Also used : SurvivorShipAlgorithmParams(org.talend.dataquality.record.linkage.grouping.swoosh.SurvivorShipAlgorithmParams) ArrayList(java.util.ArrayList) SurvivorshipKeyDefinition(org.talend.dataquality.rules.SurvivorshipKeyDefinition) SurvivorshipFunction(org.talend.dataquality.record.linkage.grouping.swoosh.SurvivorShipAlgorithmParams.SurvivorshipFunction)

Example 4 with SurvivorshipKeyDefinition

use of org.talend.dataquality.rules.SurvivorshipKeyDefinition in project tdq-studio-se by Talend.

the class AnaMatchSurvivorSection method createNewSurvivorshipKeyDefinition.

private SurvivorshipKeyDefinition createNewSurvivorshipKeyDefinition(String matchKeyName) {
    SurvivorshipKeyDefinition skd = RulesFactory.eINSTANCE.createSurvivorshipKeyDefinition();
    skd.setName(matchKeyName);
    AlgorithmDefinition createAlgorithmDefinition = RulesFactory.eINSTANCE.createAlgorithmDefinition();
    createAlgorithmDefinition.setAlgorithmType(SurvivorShipAlgorithmEnum.getTypeByIndex(0).getValue());
    // MOD TDQ-11774 set a default value for parameter
    createAlgorithmDefinition.setAlgorithmParameters(SurvivorshipUtils.DEFAULT_CONCATENATE_PARAMETER);
    skd.setFunction(createAlgorithmDefinition);
    skd.setAllowManualResolution(true);
    return skd;
}
Also used : SurvivorshipKeyDefinition(org.talend.dataquality.rules.SurvivorshipKeyDefinition) AlgorithmDefinition(org.talend.dataquality.rules.AlgorithmDefinition)

Example 5 with SurvivorshipKeyDefinition

use of org.talend.dataquality.rules.SurvivorshipKeyDefinition in project tdq-studio-se by Talend.

the class AnaMatchSurvivorSection method updateSurvivorKey.

private void updateSurvivorKey(boolean isCreateNewOne, String name, MatchKeyAndSurvivorDefinition definition) {
    EList<SurvivorshipKeyDefinition> survivorshipKeys = matchRuleDef.getSurvivorshipKeys();
    // set the survivor key: if clear, use a new one , if not clear, use the one in model
    boolean isSurvKeyFound = false;
    if (!isCreateNewOne) {
        // Use the key name to find the appropriate survivor key definition.
        for (SurvivorshipKeyDefinition survKey : survivorshipKeys) {
            if (survKey.getName().equals(name)) {
                definition.setSurvivorShipKey(survKey);
                isSurvKeyFound = true;
                break;
            }
        }
    }
    if (isCreateNewOne || !isSurvKeyFound) {
        SurvivorshipKeyDefinition survivorshipKeyDefinition = createNewSurvivorshipKeyDefinition(name);
        definition.setSurvivorShipKey(survivorshipKeyDefinition);
        survivorshipKeys.add(survivorshipKeyDefinition);
    }
}
Also used : SurvivorshipKeyDefinition(org.talend.dataquality.rules.SurvivorshipKeyDefinition)

Aggregations

SurvivorshipKeyDefinition (org.talend.dataquality.rules.SurvivorshipKeyDefinition)17 AlgorithmDefinition (org.talend.dataquality.rules.AlgorithmDefinition)4 ArrayList (java.util.ArrayList)3 ISelection (org.eclipse.jface.viewers.ISelection)3 StructuredSelection (org.eclipse.jface.viewers.StructuredSelection)3 MetadataColumn (org.talend.core.model.metadata.builder.connection.MetadataColumn)2 SurvivorShipAlgorithmParams (org.talend.dataquality.record.linkage.grouping.swoosh.SurvivorShipAlgorithmParams)2 SurvivorshipFunction (org.talend.dataquality.record.linkage.grouping.swoosh.SurvivorShipAlgorithmParams.SurvivorshipFunction)2 MatchKeyDefinition (org.talend.dataquality.rules.MatchKeyDefinition)2 MatchRuleDefinition (org.talend.dataquality.rules.MatchRuleDefinition)2 ReturnCode (org.talend.utils.sugars.ReturnCode)2 HashMap (java.util.HashMap)1 List (java.util.List)1 Map (java.util.Map)1 EList (org.eclipse.emf.common.util.EList)1 CTabItem (org.eclipse.swt.custom.CTabItem)1 GridData (org.eclipse.swt.layout.GridData)1 GridLayout (org.eclipse.swt.layout.GridLayout)1 Composite (org.eclipse.swt.widgets.Composite)1 TableItem (org.eclipse.swt.widgets.TableItem)1