Examples with IRecordMatcher org.talend.dataquality.record.linkage.record.IRecordMatcher used on opensource projects

Example 1 with IRecordMatcher

use of org.talend.dataquality.record.linkage.record.IRecordMatcher 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());
}

Example 2 with IRecordMatcher

use of org.talend.dataquality.record.linkage.record.IRecordMatcher in project tdq-studio-se by Talend.

the class AnalysisMatchParameterAdapter method getSurvivorshipAlgosMap.

/*
     * (non-Javadoc)
     * 
     * @see org.talend.dataquality.record.linkage.grouping.adapter.MatchParameterAdapter#getSurvivorshipAlgosMap(java.util.Map)
     */
@Override
public Map<IRecordMatcher, SurvivorshipFunction[]> getSurvivorshipAlgosMap(Map<Integer, SurvivorshipFunction> colIdx2DefaultSurvFunc, List<SurvivorshipFunction> survFunctions) {
    Map<IRecordMatcher, SurvivorshipFunction[]> survAlgos = new HashMap<IRecordMatcher, SurvivorshipFunction[]>();
    int matchRuleIdx = -1;
    List<List<Map<String, String>>> multiRules = analysisMatchRecordGrouping.getMultiMatchRules();
    for (List<Map<String, String>> matchrule : multiRules) {
        matchRuleIdx++;
        if (matchrule == null) {
            continue;
        }
        SurvivorshipFunction[] surFuncsInMatcher = new SurvivorshipFunction[matchrule.size()];
        int idx = 0;
        for (Map<String, String> mkDef : matchrule) {
            String matcherType = mkDef.get(IRecordGrouping.MATCHING_TYPE);
            if (AttributeMatcherType.DUMMY.name().equalsIgnoreCase(matcherType)) {
                // Find the func from default survivorship rule.
                surFuncsInMatcher[idx] = colIdx2DefaultSurvFunc.get(Integer.valueOf(mkDef.get(IRecordGrouping.COLUMN_IDX)));
                if (surFuncsInMatcher[idx] == null) {
                    // Use CONCATENATE by default if not specified .
                    surFuncsInMatcher[idx] = new SurvivorShipAlgorithmParams().new SurvivorshipFunction();
                    surFuncsInMatcher[idx].setSurvivorShipAlgoEnum(SurvivorShipAlgorithmEnum.MOST_COMMON);
                    // MOD TDQ-11774 set a default parameter
                    surFuncsInMatcher[idx].setParameter(SurvivorshipUtils.DEFAULT_CONCATENATE_PARAMETER);
                }
            } else {
                // Find the func from existing survivorship rule list.
                for (SurvivorshipFunction survFunc : survFunctions) {
                    String keyName = mkDef.get(IRecordGrouping.MATCH_KEY_NAME);
                    if (keyName.equals(survFunc.getSurvivorShipKey())) {
                        surFuncsInMatcher[idx] = survFunc;
                        break;
                    }
                }
            }
            idx++;
        }
        // Add the funcs to a specific record matcher. NOTE that the index of matcher must be coincidence to the
        // index of match rule.
        survAlgos.put(this.getCombinedRecordMatcher().getMatchers().get(matchRuleIdx), surFuncsInMatcher);
    }
    return survAlgos;
}