use of org.talend.dataquality.analysis.Analysis in project tdq-studio-se by Talend.
the class MatchAnalysisExecutorTest method testExecute.
/**
* Test method for
* {@link org.talend.dq.analysis.MatchAnalysisExecutor#execute(org.talend.dataquality.analysis.Analysis)}.
*/
@SuppressWarnings("nls")
@Test
public void testExecute() {
MatchAnalysisExecutor matchAnalysisExecutor = new MatchAnalysisExecutor();
Analysis analysis = AnalysisPackage.eINSTANCE.getAnalysisFactory().createAnalysis();
AnalysisContext context = AnalysisPackage.eINSTANCE.getAnalysisFactory().createAnalysisContext();
analysis.setContext(context);
AnalysisParameters params = AnalysisPackage.eINSTANCE.getAnalysisFactory().createAnalysisParameters();
analysis.setParameters(params);
TaggedValueHelper.setTaggedValue(analysis, TaggedValueHelper.PREVIEW_ROW_NUMBER, String.valueOf(100));
// analysisResult.setAnalysis(analysis);
context.setConnection(delimitedFileconnection);
// $NON-NLS-1$
URL fileUrl = this.getClass().getResource("match_test_data");
metadataTable = UnitTestBuildHelper.getDefault().initFileConnection(fileUrl, delimitedFileconnection);
this.name = UnitTestBuildHelper.getDefault().initColumns(context, this.metadataTable);
// Scenario 1
// - Match key: name, no block key, levenshtein attribute algorithm. groupQualityThreshold = 0.9d, matchInterval
// = 0.95d .
double groupQualityThreshold = 0.9d;
double matchInterval = 0.95d;
assertScenario1(matchAnalysisExecutor, analysis, name, "name", groupQualityThreshold, matchInterval);
// Scenario 2
// - Same to scenario 1, EXCEPT matchInterval = 0.8d .
matchInterval = 0.8d;
assertScenario2(matchAnalysisExecutor, analysis, name, "name", groupQualityThreshold, matchInterval);
// Scenario 3
// - Same to scenario 2, EXCEPT groupQualityThreshold = 0.95d.
groupQualityThreshold = 0.95d;
assertScenario3(matchAnalysisExecutor, analysis, name, "name", groupQualityThreshold, matchInterval);
// Scenario 4
// - Same to scenario 3, EXCEPT a new blocking key = country.
assertScenario4(matchAnalysisExecutor, analysis, name, "name", groupQualityThreshold, matchInterval);
}
use of org.talend.dataquality.analysis.Analysis in project tdq-studio-se by Talend.
the class TdqAnalysisConnectionPoolTest method setUp.
/**
* DOC yyin Comment method "setUp".
*
* @throws java.lang.Exception
*/
@Before
public void setUp() throws Exception {
analysis = mock(Analysis.class);
AnalysisContext context = mock(AnalysisContext.class);
AnalysisResult result = mock(AnalysisResult.class);
dataManager = mock(org.talend.core.model.metadata.builder.connection.Connection.class);
ExecutionInformations resultMetadata = mock(ExecutionInformations.class);
PowerMockito.mockStatic(AnalysisHandler.class);
AnalysisHandler mockHandler = mock(AnalysisHandler.class);
when(AnalysisHandler.createHandler(analysis)).thenReturn(mockHandler);
when(mockHandler.getNumberOfConnectionsPerAnalysis()).thenReturn(5);
when(context.getConnection()).thenReturn(dataManager);
when(analysis.getContext()).thenReturn(context);
when(analysis.getResults()).thenReturn(result);
when(result.getResultMetadata()).thenReturn(resultMetadata);
connPool = new TdqAnalysisConnectionPool(analysis, 5);
}
use of org.talend.dataquality.analysis.Analysis in project tdq-studio-se by Talend.
the class PatternExplorerRealTest method setUp.
/**
* DOC zshen Comment method "setUp".
*
* @throws java.lang.Exception
*/
@Before
public void setUp() throws Exception {
patternExplorer = new RegexPatternExplorer();
// mock setEntity
PatternMatchingIndicator indicator = IndicatorsFactoryImpl.eINSTANCE.createRegexpMatchingIndicator();
// create pattern parameter
IndicatorParameters createIndicatorParameters = IndicatorsFactory.eINSTANCE.createIndicatorParameters();
indicator.setParameters(createIndicatorParameters);
Domain createDomain = DomainFactoryImpl.eINSTANCE.createDomain();
createIndicatorParameters.setDataValidDomain(createDomain);
Pattern createPattern = PatternFactoryImpl.eINSTANCE.createPattern();
createDomain.getPatterns().add(createPattern);
RegularExpression createPatternComponent = PatternFactoryImpl.eINSTANCE.createRegularExpression();
createPattern.getComponents().add(createPatternComponent);
TdExpression createTdExpression = RelationalFactoryImpl.eINSTANCE.createTdExpression();
createPatternComponent.setExpression(createTdExpression);
createTdExpression.setLanguage(SupportDBUrlType.HIVEDEFAULTURL.getLanguage());
// $NON-NLS-1$
createTdExpression.setBody("`su.*`");
ModelElement element = RelationalFactoryImpl.eINSTANCE.createTdColumn();
// $NON-NLS-1$
element.setName("lname");
indicator.setAnalyzedElement(element);
Expression createIndiExpression = CoreFactoryImpl.eINSTANCE.createExpression();
createIndiExpression.setLanguage(SupportDBUrlType.HIVEDEFAULTURL.getLanguage());
// $NON-NLS-1$
createIndiExpression.setBody("SELECT * FROM `tbi`.`customer` WHERE (customer.lname = \"sunny\")");
indicator.setInstantiatedExpression(createIndiExpression);
Analysis analysis = AnalysisFactoryImpl.eINSTANCE.createAnalysis();
AnalysisContext createAnalysisContext = AnalysisFactoryImpl.eINSTANCE.createAnalysisContext();
analysis.setContext(createAnalysisContext);
DatabaseConnection createDatabaseConnection = ConnectionFactoryImpl.eINSTANCE.createDatabaseConnection();
TaggedValueHelper.setTaggedValue(createDatabaseConnection, TaggedValueHelper.DB_PRODUCT_NAME, SupportDBUrlType.HIVEDEFAULTURL.getLanguage());
// $NON-NLS-1$
TaggedValueHelper.setTaggedValue(createDatabaseConnection, TaggedValueHelper.DB_PRODUCT_VERSION, "1.0");
// $NON-NLS-1$
TaggedValueHelper.setTaggedValue(createDatabaseConnection, TaggedValueHelper.DB_IDENTIFIER_QUOTE_STRING, "`");
createAnalysisContext.setConnection(createDatabaseConnection);
patternExplorer.setAnalysis(analysis);
ChartDataEntity cdEntity = new ChartDataEntity();
cdEntity.setIndicator(indicator);
patternExplorer.setEnitty(cdEntity);
}
use of org.talend.dataquality.analysis.Analysis in project tdq-studio-se by Talend.
the class PatternExplorerRealTest method testGetValidValuesStatement_2.
/**
* Test method for {@link org.talend.dq.analysis.explore.PatternExplorer#getValidValuesStatement()}. when the test
* for indicator is user define indicator
*/
@Test
public void testGetValidValuesStatement_2() {
patternExplorer = new RegexPatternExplorer();
// mock setEntity
PatternMatchingIndicator indicator = IndicatorsFactoryImpl.eINSTANCE.createRegexpMatchingIndicator();
UDIndicatorDefinition indicatorDef = UserdefineFactoryImpl.eINSTANCE.createUDIndicatorDefinition();
indicator.setIndicatorDefinition(indicatorDef);
TdExpression udiTdExpression = RelationalFactoryImpl.eINSTANCE.createTdExpression();
udiTdExpression.setLanguage(SupportDBUrlType.HIVEDEFAULTURL.getLanguage());
udiTdExpression.setBody(ViewValidValues);
indicatorDef.getViewValidValuesExpression().add(udiTdExpression);
// create pattern parameter
IndicatorParameters createIndicatorParameters = IndicatorsFactory.eINSTANCE.createIndicatorParameters();
indicator.setParameters(createIndicatorParameters);
Domain createDomain = DomainFactoryImpl.eINSTANCE.createDomain();
createIndicatorParameters.setDataValidDomain(createDomain);
Pattern createPattern = PatternFactoryImpl.eINSTANCE.createPattern();
createDomain.getPatterns().add(createPattern);
RegularExpression createPatternComponent = PatternFactoryImpl.eINSTANCE.createRegularExpression();
createPattern.getComponents().add(createPatternComponent);
TdExpression createTdExpression = RelationalFactoryImpl.eINSTANCE.createTdExpression();
createPatternComponent.setExpression(createTdExpression);
createTdExpression.setLanguage(SupportDBUrlType.HIVEDEFAULTURL.getLanguage());
// $NON-NLS-1$
createTdExpression.setBody("`su.*`");
TdColumn element = RelationalFactoryImpl.eINSTANCE.createTdColumn();
// $NON-NLS-1$
element.setName("lname");
TdTable createTdTable = RelationalFactoryImpl.eINSTANCE.createTdTable();
// $NON-NLS-1$
createTdTable.setName("table1");
TableHelper.addColumn(createTdTable, element);
Catalog createCatalog = orgomg.cwm.resource.relational.impl.RelationalFactoryImpl.eINSTANCE.createCatalog();
// $NON-NLS-1$
createCatalog.setName("catalog1");
List<TdTable> tableList = new ArrayList<TdTable>();
tableList.add(createTdTable);
CatalogHelper.addTables(tableList, createCatalog);
indicator.setAnalyzedElement(element);
Expression createIndiExpression = CoreFactoryImpl.eINSTANCE.createExpression();
createIndiExpression.setLanguage(SupportDBUrlType.HIVEDEFAULTURL.getLanguage());
// $NON-NLS-1$
createIndiExpression.setBody("SELECT * FROM `tbi`.`customer` WHERE (customer.lname = \"sunny\")");
indicator.setInstantiatedExpression(createIndiExpression);
Analysis analysis = AnalysisFactoryImpl.eINSTANCE.createAnalysis();
AnalysisContext createAnalysisContext = AnalysisFactoryImpl.eINSTANCE.createAnalysisContext();
analysis.setContext(createAnalysisContext);
DatabaseConnection createDatabaseConnection = ConnectionFactoryImpl.eINSTANCE.createDatabaseConnection();
TaggedValueHelper.setTaggedValue(createDatabaseConnection, TaggedValueHelper.DB_PRODUCT_NAME, SupportDBUrlType.HIVEDEFAULTURL.getLanguage());
// $NON-NLS-1$
TaggedValueHelper.setTaggedValue(createDatabaseConnection, TaggedValueHelper.DB_PRODUCT_VERSION, "1.0");
// $NON-NLS-1$
TaggedValueHelper.setTaggedValue(createDatabaseConnection, TaggedValueHelper.DB_IDENTIFIER_QUOTE_STRING, "`");
createAnalysisContext.setConnection(createDatabaseConnection);
patternExplorer.setAnalysis(analysis);
ChartDataEntity cdEntity = new ChartDataEntity();
cdEntity.setIndicator(indicator);
patternExplorer.setEnitty(cdEntity);
Assert.assertTrue(patternExplorer.setAnalysis(analysis));
String clause = patternExplorer.getValidValuesStatement();
// $NON-NLS-1$
assertEquals("SELECT `lname` FROM `catalog1`.`table1` ", clause);
}
use of org.talend.dataquality.analysis.Analysis in project tdq-studio-se by Talend.
the class PatternExplorerRealTest method testGetInvalidRowsStatement_4.
/**
* Test method for {@link org.talend.dq.analysis.explore.PatternExplorer#getInvalidRowsStatement()}. when the test
* for indicator is user define indicator
*/
@Test
public void testGetInvalidRowsStatement_4() {
patternExplorer = new RegexPatternExplorer();
// mock setEntity
PatternMatchingIndicator indicator = IndicatorsFactoryImpl.eINSTANCE.createRegexpMatchingIndicator();
UDIndicatorDefinition indicatorDef = UserdefineFactoryImpl.eINSTANCE.createUDIndicatorDefinition();
indicator.setIndicatorDefinition(indicatorDef);
TdExpression udiTdExpression = RelationalFactoryImpl.eINSTANCE.createTdExpression();
udiTdExpression.setLanguage(SupportDBUrlType.HIVEDEFAULTURL.getLanguage());
udiTdExpression.setBody(ViewInvalidRows);
indicatorDef.getViewInvalidRowsExpression().add(udiTdExpression);
// create pattern parameter
IndicatorParameters createIndicatorParameters = IndicatorsFactory.eINSTANCE.createIndicatorParameters();
indicator.setParameters(createIndicatorParameters);
Domain createDomain = DomainFactoryImpl.eINSTANCE.createDomain();
createIndicatorParameters.setDataValidDomain(createDomain);
Pattern createPattern = PatternFactoryImpl.eINSTANCE.createPattern();
createDomain.getPatterns().add(createPattern);
RegularExpression createPatternComponent = PatternFactoryImpl.eINSTANCE.createRegularExpression();
createPattern.getComponents().add(createPatternComponent);
TdExpression createTdExpression = RelationalFactoryImpl.eINSTANCE.createTdExpression();
createPatternComponent.setExpression(createTdExpression);
createTdExpression.setLanguage(SupportDBUrlType.HIVEDEFAULTURL.getLanguage());
// $NON-NLS-1$
createTdExpression.setBody("`su.*`");
TdColumn element = RelationalFactoryImpl.eINSTANCE.createTdColumn();
// $NON-NLS-1$
element.setName("lname");
TdTable createTdTable = RelationalFactoryImpl.eINSTANCE.createTdTable();
// $NON-NLS-1$
createTdTable.setName("table1");
TableHelper.addColumn(createTdTable, element);
Catalog createCatalog = orgomg.cwm.resource.relational.impl.RelationalFactoryImpl.eINSTANCE.createCatalog();
// $NON-NLS-1$
createCatalog.setName("catalog1");
List<TdTable> tableList = new ArrayList<TdTable>();
tableList.add(createTdTable);
CatalogHelper.addTables(tableList, createCatalog);
indicator.setAnalyzedElement(element);
Expression createIndiExpression = CoreFactoryImpl.eINSTANCE.createExpression();
createIndiExpression.setLanguage(SupportDBUrlType.HIVEDEFAULTURL.getLanguage());
// $NON-NLS-1$
createIndiExpression.setBody("SELECT * FROM `tbi`.`customer` WHERE (customer.lname = \"sunny\")");
indicator.setInstantiatedExpression(createIndiExpression);
Analysis analysis = AnalysisFactoryImpl.eINSTANCE.createAnalysis();
AnalysisContext createAnalysisContext = AnalysisFactoryImpl.eINSTANCE.createAnalysisContext();
analysis.setContext(createAnalysisContext);
DatabaseConnection createDatabaseConnection = ConnectionFactoryImpl.eINSTANCE.createDatabaseConnection();
TaggedValueHelper.setTaggedValue(createDatabaseConnection, TaggedValueHelper.DB_PRODUCT_NAME, SupportDBUrlType.HIVEDEFAULTURL.getLanguage());
// $NON-NLS-1$
TaggedValueHelper.setTaggedValue(createDatabaseConnection, TaggedValueHelper.DB_PRODUCT_VERSION, "1.0");
// $NON-NLS-1$
TaggedValueHelper.setTaggedValue(createDatabaseConnection, TaggedValueHelper.DB_IDENTIFIER_QUOTE_STRING, "`");
createAnalysisContext.setConnection(createDatabaseConnection);
ChartDataEntity cdEntity = new ChartDataEntity();
cdEntity.setIndicator(indicator);
Assert.assertTrue(patternExplorer.setAnalysis(analysis));
patternExplorer.setEnitty(cdEntity);
String clause = patternExplorer.getInvalidRowsStatement();
// $NON-NLS-1$
assertEquals("SELECT * FROM `catalog1`.`table1` WHERE NOT (id>=1) ", clause);
}
Aggregations