Examples with BlockKeyIndicator org.talend.dataquality.indicators.columnset.BlockKeyIndicator used on opensource projects

Example 1 with BlockKeyIndicator

use of org.talend.dataquality.indicators.columnset.BlockKeyIndicator in project tdq-studio-se by Talend.

the class ExecuteMatchRuleHandlerTest method testExecute4.

/**
 * Test method for
 * {@link org.talend.dq.analysis.ExecuteMatchRuleHandler#execute(java.util.Map, org.talend.dataquality.indicators.columnset.RecordMatchingIndicator, java.util.List, org.talend.dataquality.indicators.columnset.BlockKeyIndicator)}
 * .
 *
 * same to case 3 but the match rule order is exchange
 */
@Test
public void testExecute4() {
    Map<MetadataColumn, String> columnMap = new HashMap<MetadataColumn, String>();
    MetadataColumn col0 = ConnectionFactory.eINSTANCE.createMetadataColumn();
    col0.setName(columnName0);
    // $NON-NLS-1$
    columnMap.put(col0, "0");
    MetadataColumn col1 = ConnectionFactory.eINSTANCE.createMetadataColumn();
    col1.setName(columnName1);
    // $NON-NLS-1$
    columnMap.put(col1, "1");
    MetadataColumn col2 = ConnectionFactory.eINSTANCE.createMetadataColumn();
    col2.setName(columnName2);
    // $NON-NLS-1$
    columnMap.put(col2, "2");
    MetadataColumn col3 = ConnectionFactory.eINSTANCE.createMetadataColumn();
    col3.setName(columnName3);
    // $NON-NLS-1$
    columnMap.put(col3, "3");
    RecordMatchingIndicator recordMatchingIndicator = ColumnsetFactory.eINSTANCE.createRecordMatchingIndicator();
    MatchRuleDefinition matchRuleDef = RulesPackage.eINSTANCE.getRulesFactory().createMatchRuleDefinition();
    recordMatchingIndicator.setBuiltInMatchRuleDefinition(matchRuleDef);
    // create match rule
    MatchRule matchRule2 = RulesFactory.eINSTANCE.createMatchRule();
    MatchKeyDefinition createMatchKeyDefinition2 = RulesFactory.eINSTANCE.createMatchKeyDefinition();
    matchRule2.getMatchKeys().add(createMatchKeyDefinition2);
    createMatchKeyDefinition2.setColumn(columnName3);
    createMatchKeyDefinition2.setConfidenceWeight(1);
    // $NON-NLS-1$
    createMatchKeyDefinition2.setName("rule1.matchkey1");
    createMatchKeyDefinition2.setHandleNull(HandleNullEnum.NULL_MATCH_NULL.getValue());
    AlgorithmDefinition createAlgorithmDefinition2 = RulesFactory.eINSTANCE.createAlgorithmDefinition();
    createAlgorithmDefinition2.setAlgorithmType(AttributeMatcherType.EXACT.name());
    createMatchKeyDefinition2.setAlgorithm(createAlgorithmDefinition2);
    matchRuleDef.getMatchRules().add(matchRule2);
    // create match rule
    MatchRule matchRule1 = RulesFactory.eINSTANCE.createMatchRule();
    MatchKeyDefinition createMatchKeyDefinition1 = RulesFactory.eINSTANCE.createMatchKeyDefinition();
    matchRule1.getMatchKeys().add(createMatchKeyDefinition1);
    createMatchKeyDefinition1.setColumn(columnName2);
    createMatchKeyDefinition1.setConfidenceWeight(1);
    // $NON-NLS-1$
    createMatchKeyDefinition1.setName("rule1.matchkey1");
    createMatchKeyDefinition1.setHandleNull(HandleNullEnum.NULL_MATCH_NULL.getValue());
    AlgorithmDefinition createAlgorithmDefinition1 = RulesFactory.eINSTANCE.createAlgorithmDefinition();
    createAlgorithmDefinition1.setAlgorithmType(AttributeMatcherType.EXACT.name());
    createMatchKeyDefinition1.setAlgorithm(createAlgorithmDefinition1);
    matchRuleDef.getMatchRules().add(matchRule1);
    // create block key
    BlockKeyDefinition createBlockKeyDefinition = RulesFactory.eINSTANCE.createBlockKeyDefinition();
    createBlockKeyDefinition.setColumn(columnName1);
    // $NON-NLS-1$
    createBlockKeyDefinition.setName("blockKey1");
    // setPreAlgorithm
    AlgorithmDefinition blockPreAlgorithm = RulesFactory.eINSTANCE.createAlgorithmDefinition();
    blockPreAlgorithm.setAlgorithmType(BlockingKeyPreAlgorithmEnum.NON_ALGO.getValue());
    createBlockKeyDefinition.setPreAlgorithm(blockPreAlgorithm);
    // setAlgorithm
    AlgorithmDefinition blockAlgorithm = RulesFactory.eINSTANCE.createAlgorithmDefinition();
    blockAlgorithm.setAlgorithmType(BlockingKeyAlgorithmEnum.EXACT.getValue());
    createBlockKeyDefinition.setAlgorithm(blockAlgorithm);
    // setPostAlgorithm
    AlgorithmDefinition blockPostAlgorithm = RulesFactory.eINSTANCE.createAlgorithmDefinition();
    blockPostAlgorithm.setAlgorithmType(BlockingKeyPostAlgorithmEnum.NON_ALGO.getValue());
    createBlockKeyDefinition.setPostAlgorithm(blockPostAlgorithm);
    matchRuleDef.getBlockKeys().add(createBlockKeyDefinition);
    List<Object[]> matchRows = new ArrayList<Object[]>();
    // $NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$ //$NON-NLS-4$
    matchRows.add(new String[] { "id1", "name1", "number1", "date1" });
    // $NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$ //$NON-NLS-4$
    matchRows.add(new String[] { "id2", "name1", "number2", "date1" });
    // $NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$ //$NON-NLS-4$
    matchRows.add(new String[] { "id3", "name2", "number2", "date3" });
    // $NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$ //$NON-NLS-4$
    matchRows.add(new String[] { "id4", "name2", "number2", "date1" });
    BlockKeyIndicator blockKeyIndicator = ColumnsetFactory.eINSTANCE.createBlockKeyIndicator();
    ExecuteMatchRuleHandler execHandler = new ExecuteMatchRuleHandler();
    MatchGroupResultConsumer matchResultConsumer = createMatchGroupResultConsumer(columnMap, recordMatchingIndicator);
    TypedReturnCode<MatchGroupResultConsumer> executeResult = execHandler.execute(columnMap, recordMatchingIndicator, matchRows, blockKeyIndicator, matchResultConsumer);
    Assert.assertTrue(executeResult.isOk());
    Assert.assertTrue(executeResult.getMessage() == null);
    Assert.assertTrue(executeResult.getObject() != null);
    MatchGroupResultConsumer ResultConsumer = executeResult.getObject();
    List<Object[]> fullMatchResult = ResultConsumer.getFullMatchResult();
    Assert.assertTrue(fullMatchResult.size() == 4);
    for (int i = 0; i < fullMatchResult.size(); i++) {
        Object[] objectArray = fullMatchResult.get(i);
        Object masterValue = objectArray[7];
        Object idValue = objectArray[0];
        // id2 is because of matchRule1 id4 is because of matchRule2
        if ("id2".equals(idValue) || "id4".equals(idValue)) {
            // $NON-NLS-1$ //$NON-NLS-2$
            Assert.assertFalse(Boolean.parseBoolean(masterValue.toString()));
        } else {
            Assert.assertTrue(Boolean.parseBoolean(masterValue.toString()));
        }
    }
}

Example 2 with BlockKeyIndicator

use of org.talend.dataquality.indicators.columnset.BlockKeyIndicator in project tdq-studio-se by Talend.

the class ExecuteMatchRuleHandlerTest method testExecute1.

/**
 * Test method for
 * {@link org.talend.dq.analysis.ExecuteMatchRuleHandler#execute(java.util.Map, org.talend.dataquality.indicators.columnset.RecordMatchingIndicator, java.util.List, org.talend.dataquality.indicators.columnset.BlockKeyIndicator)}
 * .
 *
 * no block key one match key
 */
@Test
public void testExecute1() {
    Map<MetadataColumn, String> columnMap = new HashMap<MetadataColumn, String>();
    MetadataColumn col0 = ConnectionFactory.eINSTANCE.createMetadataColumn();
    col0.setName(columnName0);
    // $NON-NLS-1$
    columnMap.put(col0, "0");
    MetadataColumn col1 = ConnectionFactory.eINSTANCE.createMetadataColumn();
    col1.setName(columnName1);
    // $NON-NLS-1$
    columnMap.put(col1, "1");
    MetadataColumn col2 = ConnectionFactory.eINSTANCE.createMetadataColumn();
    col2.setName(columnName2);
    // $NON-NLS-1$
    columnMap.put(col2, "2");
    MetadataColumn col3 = ConnectionFactory.eINSTANCE.createMetadataColumn();
    col3.setName(columnName3);
    // $NON-NLS-1$
    columnMap.put(col3, "3");
    // create match key
    RecordMatchingIndicator recordMatchingIndicator = ColumnsetFactory.eINSTANCE.createRecordMatchingIndicator();
    MatchRuleDefinition matchRuleDef = RulesPackage.eINSTANCE.getRulesFactory().createMatchRuleDefinition();
    recordMatchingIndicator.setBuiltInMatchRuleDefinition(matchRuleDef);
    MatchRule createMatchRule1 = RulesFactory.eINSTANCE.createMatchRule();
    MatchKeyDefinition createMatchKeyDefinition1 = RulesFactory.eINSTANCE.createMatchKeyDefinition();
    createMatchRule1.getMatchKeys().add(createMatchKeyDefinition1);
    createMatchKeyDefinition1.setColumn(columnName0);
    createMatchKeyDefinition1.setConfidenceWeight(1);
    // $NON-NLS-1$
    createMatchKeyDefinition1.setName("rule1.matchkey1");
    createMatchKeyDefinition1.setHandleNull(HandleNullEnum.NULL_MATCH_NULL.getValue());
    AlgorithmDefinition createAlgorithmDefinition1 = RulesFactory.eINSTANCE.createAlgorithmDefinition();
    createAlgorithmDefinition1.setAlgorithmType(AttributeMatcherType.EXACT.name());
    createMatchKeyDefinition1.setAlgorithm(createAlgorithmDefinition1);
    matchRuleDef.getMatchRules().add(createMatchRule1);
    // input data
    List<Object[]> matchRows = new ArrayList<Object[]>();
    // $NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$ //$NON-NLS-4$
    matchRows.add(new String[] { "id1", "name1", "number1", "date1" });
    // $NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$ //$NON-NLS-4$
    matchRows.add(new String[] { "id2", "name2", "number2", "date2" });
    // $NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$ //$NON-NLS-4$
    matchRows.add(new String[] { "id3", "name1", "number3", "date3" });
    // $NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$ //$NON-NLS-4$
    matchRows.add(new String[] { "id4", "name4", "number2", "date1" });
    BlockKeyIndicator blockKeyIndicator = ColumnsetFactory.eINSTANCE.createBlockKeyIndicator();
    ExecuteMatchRuleHandler execHandler = new ExecuteMatchRuleHandler();
    MatchGroupResultConsumer matchResultConsumer = createMatchGroupResultConsumer(columnMap, recordMatchingIndicator);
    TypedReturnCode<MatchGroupResultConsumer> executeResult = execHandler.execute(columnMap, recordMatchingIndicator, matchRows, blockKeyIndicator, matchResultConsumer);
    Assert.assertTrue(executeResult.isOk());
    Assert.assertTrue(executeResult.getMessage() == null);
    Assert.assertTrue(executeResult.getObject() != null);
    MatchGroupResultConsumer ResultConsumer = executeResult.getObject();
    List<Object[]> fullMatchResult = ResultConsumer.getFullMatchResult();
    Assert.assertTrue(fullMatchResult.size() == 4);
    // every input data is master data
    for (Object[] objectArray : fullMatchResult) {
        Object object = objectArray[7];
        Assert.assertTrue(Boolean.parseBoolean(object.toString()));
    }
}

Example 3 with BlockKeyIndicator

use of org.talend.dataquality.indicators.columnset.BlockKeyIndicator in project tdq-studio-se by Talend.

the class MatchAnalysisExecutorTest method executeAnalysis.

/**
 * DOC zhao Comment method "executeAnalysis".
 *
 * @param matchAnalysisExecutor
 * @param analysis
 * @param matchIndicator
 */
private void executeAnalysis(MatchAnalysisExecutor matchAnalysisExecutor, Analysis analysis, RecordMatchingIndicator matchIndicator) {
    BlockKeyIndicator blockKeyIndicator = ColumnsetPackage.eINSTANCE.getColumnsetFactory().createBlockKeyIndicator();
    AnalysisResult anaResult = AnalysisPackage.eINSTANCE.getAnalysisFactory().createAnalysisResult();
    anaResult.setResultMetadata(AnalysisPackage.eINSTANCE.getAnalysisFactory().createExecutionInformations());
    analysis.setResults(anaResult);
    analysis.getResults().getIndicators().add(matchIndicator);
    analysis.getResults().getIndicators().add(blockKeyIndicator);
    matchAnalysisExecutor.setMonitor(new NullProgressMonitor());
    matchAnalysisExecutor.execute(analysis);
}

Example 4 with BlockKeyIndicator

use of org.talend.dataquality.indicators.columnset.BlockKeyIndicator in project tdq-studio-se by Talend.

the class MatchAnalysisExecutor method execute.

/*
     * (non-Javadoc)
     * 
     * @see org.talend.dq.analysis.IAnalysisExecutor#execute(org.talend.dataquality.analysis.Analysis)
     */
public ReturnCode execute(Analysis analysis) {
    assert analysis != null;
    // --- preconditions
    ReturnCode rc = AnalysisExecutorHelper.check(analysis);
    if (!rc.isOk()) {
        AnalysisExecutorHelper.setExecutionInfoInAnalysisResult(analysis, rc.isOk(), rc.getMessage());
        return rc;
    }
    // --- creation time
    final long startime = AnalysisExecutorHelper.setExecutionDateInAnalysisResult(analysis);
    EList<Indicator> indicators = analysis.getResults().getIndicators();
    RecordMatchingIndicator recordMatchingIndicator = null;
    BlockKeyIndicator blockKeyIndicator = null;
    for (Indicator ind : indicators) {
        if (ind instanceof RecordMatchingIndicator) {
            recordMatchingIndicator = (RecordMatchingIndicator) ind;
        } else if (ind instanceof BlockKeyIndicator) {
            blockKeyIndicator = (BlockKeyIndicator) ind;
        }
    }
    if (recordMatchingIndicator == null || blockKeyIndicator == null) {
        rc.setOk(Boolean.FALSE);
        // $NON-NLS-1$
        rc.setMessage(Messages.getString("MatchAnalysisExecutor.noIndicators"));
        AnalysisExecutorHelper.setExecutionInfoInAnalysisResult(analysis, rc.isOk(), rc.getMessage());
        return rc;
    }
    List<ModelElement> anlayzedElements = analysis.getContext().getAnalysedElements();
    if (anlayzedElements == null || anlayzedElements.size() == 0) {
        rc.setOk(Boolean.FALSE);
        // $NON-NLS-1$
        rc.setMessage(Messages.getString("MatchAnalysisExecutor.EmptyAnalyzedElement"));
        AnalysisExecutorHelper.setExecutionInfoInAnalysisResult(analysis, rc.isOk(), rc.getMessage());
        return rc;
    }
    // TDQ-9664 msjian: check the store on disk path.
    Boolean isStoreOnDisk = TaggedValueHelper.getValueBoolean(SQLExecutor.STORE_ON_DISK_KEY, analysis);
    if (isStoreOnDisk) {
        String tempDataPath = TaggedValueHelper.getValueString(SQLExecutor.TEMP_DATA_DIR, analysis);
        File file = new File(tempDataPath);
        if (!file.exists() || !file.isDirectory()) {
            rc.setOk(Boolean.FALSE);
            // $NON-NLS-1$
            rc.setMessage(Messages.getString("MatchAnalysisExecutor.InvalidPath", file.getPath()));
            AnalysisExecutorHelper.setExecutionInfoInAnalysisResult(analysis, rc.isOk(), rc.getMessage());
            return rc;
        }
    }
    // TDQ-9664~
    Map<MetadataColumn, String> columnMap = getColumn2IndexMap(anlayzedElements);
    ISQLExecutor sqlExecutor = getSQLExectutor(analysis, recordMatchingIndicator, columnMap);
    if (sqlExecutor == null) {
        rc.setOk(Boolean.FALSE);
        // $NON-NLS-1$
        rc.setMessage(Messages.getString("MatchAnalysisExecutor.noSqlExecutor"));
        AnalysisExecutorHelper.setExecutionInfoInAnalysisResult(analysis, rc.isOk(), rc.getMessage());
        return rc;
    }
    if (getMonitor() != null) {
        getMonitor().worked(20);
    }
    // Set schema for match key.
    TypedReturnCode<MatchGroupResultConsumer> returnCode = new TypedReturnCode<MatchGroupResultConsumer>();
    MetadataColumn[] completeColumnSchema = AnalysisRecordGroupingUtils.getCompleteColumnSchema(columnMap);
    String[] colSchemaString = new String[completeColumnSchema.length];
    int idx = 0;
    for (MetadataColumn metadataCol : completeColumnSchema) {
        colSchemaString[idx++] = metadataCol.getName();
    }
    recordMatchingIndicator.setMatchRowSchema(colSchemaString);
    recordMatchingIndicator.reset();
    MatchGroupResultConsumer matchResultConsumer = createMatchGroupResultConsumer(recordMatchingIndicator);
    if (sqlExecutor.isStoreOnDisk()) {
        // need to execute the query
        try {
            sqlExecutor.executeQuery(analysis.getContext().getConnection(), analysis.getContext().getAnalysedElements());
        } catch (SQLException e) {
            log.error(e, e);
            rc.setOk(Boolean.FALSE);
            rc.setMessage(e.getMessage());
            AnalysisExecutorHelper.setExecutionInfoInAnalysisResult(analysis, rc.isOk(), rc.getMessage());
            return rc;
        }
        try {
            TypedReturnCode<Object> result = StoreOnDiskUtils.getDefault().executeWithStoreOnDisk(columnMap, recordMatchingIndicator, blockKeyIndicator, sqlExecutor.getStoreOnDiskHandler(), matchResultConsumer);
            if (result != null) {
                returnCode.setObject((MatchGroupResultConsumer) result.getObject());
                returnCode.setOk(result.isOk());
                returnCode.setMessage(result.getMessage());
            }
        } catch (Exception e) {
            log.error(e, e);
            returnCode.setMessage(e.getMessage());
            returnCode.setOk(false);
        }
    } else {
        // Added TDQ-9320 , use the result set iterator to replace the list of result in the memory.
        try {
            Iterator<Record> resultSetIterator = sqlExecutor.getResultSetIterator(analysis.getContext().getConnection(), anlayzedElements);
            BlockAndMatchManager bAndmManager = new BlockAndMatchManager(resultSetIterator, matchResultConsumer, columnMap, recordMatchingIndicator, blockKeyIndicator);
            bAndmManager.run();
        } catch (SQLException e) {
            log.error(e, e);
            rc.setOk(Boolean.FALSE);
            rc.setMessage(e.getMessage());
            AnalysisExecutorHelper.setExecutionInfoInAnalysisResult(analysis, rc.isOk(), rc.getMessage());
            return rc;
        } catch (BusinessException e) {
            log.error(e, e);
            rc.setOk(Boolean.FALSE);
            rc.setMessage(e.getMessage());
            AnalysisExecutorHelper.setExecutionInfoInAnalysisResult(analysis, rc.isOk(), rc.getMessage());
            return rc;
        }
    }
    if (!returnCode.isOk()) {
        rc.setOk(returnCode.isOk());
        rc.setMessage(returnCode.getMessage());
    }
    if (getMonitor() != null) {
        getMonitor().worked(20);
    }
    if (isLowMemory) {
        // $NON-NLS-1$
        rc.setMessage(Messages.getString("Evaluator.OutOfMomory", usedMemory));
    }
    // nodify the master page
    refreshTableWithMatchFullResult(analysis);
    // --- set metadata information of analysis
    AnalysisExecutorHelper.setExecutionInfoInAnalysisResult(analysis, rc.isOk(), rc.getMessage());
    // --- compute execution duration
    if (this.continueRun()) {
        long endtime = System.currentTimeMillis();
        final ExecutionInformations resultMetadata = analysis.getResults().getResultMetadata();
        resultMetadata.setExecutionDuration((int) (endtime - startime));
        resultMetadata.setOutThreshold(false);
    }
    if (getMonitor() != null) {
        getMonitor().worked(20);
    }
    return rc;
}

Example 5 with BlockKeyIndicator

use of org.talend.dataquality.indicators.columnset.BlockKeyIndicator in project tdq-studio-se by Talend.

the class MatchRuleAnlaysisUtils method getNeedIndicatorFromAna.

/**
 * Get recording matching indicator and Blocking Indicator from analysis
 *
 * @param analysis
 * @return the index 0 will be RecordMatchingIndicator and index 1 will be BlockKeyIndicator
 */
public static Object[] getNeedIndicatorFromAna(Analysis analysis) {
    Object[] returnList = new Object[2];
    EList<Indicator> indicators = analysis.getResults().getIndicators();
    for (Indicator ind : indicators) {
        if (ind instanceof RecordMatchingIndicator) {
            returnList[0] = ind;
        } else if (ind instanceof BlockKeyIndicator) {
            returnList[1] = ind;
        }
    }
    // If match rule definition is null, create a default.
    if (returnList[0] == null) {
        returnList[0] = ColumnsetPackage.eINSTANCE.getColumnsetFactory().createRecordMatchingIndicator();
    }
    // If blocking key indicator is nul, create a default.
    if (returnList[1] == null) {
        returnList[1] = ColumnsetPackage.eINSTANCE.getColumnsetFactory().createBlockKeyIndicator();
    }
    return returnList;
}