Search in sources :

Example 1 with SumIndicator

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

the class IndicatorEvaluationMain method main.

/**
 * DOC scorreia Comment method "main".
 *
 * @param args
 */
public static void main(String[] args) {
    TypedProperties connectionParams = PropertiesLoader.getProperties(IndicatorEvaluator.class, "db.properties");
    String driverClassName = connectionParams.getProperty("driver");
    String dbUrl = connectionParams.getProperty("url");
    try {
        TimeTracer tt = new TimeTracer("Indicator evaluation", null);
        tt.start();
        // create connection
        Connection connection = ConnectionUtils.createConnection(dbUrl, driverClassName, connectionParams);
        String database = "test";
        String tableName = "my_test";
        // --- columns to analyze
        String[] columnsArray = new String[] { // 0
        "my_int", // 1
        "my_double", // 2
        "my_text", // 4
        "my_date", // 3
        "my_string", // 5
        "my_int_null" };
        List<String> columns = Arrays.asList(columnsArray);
        // store in file
        File file = new File("out/columnTest_0.1.ana");
        EMFUtil util = new EMFUtil();
        Resource resource = util.getResourceSet().createResource(URI.createFileURI(file.getAbsolutePath()));
        rContents = resource.getContents();
        evaluator.setConnection(connection);
        // --- create indicators
        RowCountIndicator rowCountIndicator = IndicatorsFactory.eINSTANCE.createRowCountIndicator();
        NullCountIndicator nullCountIndicator = IndicatorsFactory.eINSTANCE.createNullCountIndicator();
        DistinctCountIndicator distinctCountIndicator = IndicatorsFactory.eINSTANCE.createDistinctCountIndicator();
        DistinctCountIndicator distinctCountIndicator2 = IndicatorsFactory.eINSTANCE.createDistinctCountIndicator();
        UniqueCountIndicator uniqueCountIndicator = IndicatorsFactory.eINSTANCE.createUniqueCountIndicator();
        DuplicateCountIndicator duplicateCountIndicator = IndicatorsFactory.eINSTANCE.createDuplicateCountIndicator();
        BlankCountIndicator blankCountIndicator = IndicatorsFactory.eINSTANCE.createBlankCountIndicator();
        MinLengthIndicator minLengthIndicator = IndicatorsFactory.eINSTANCE.createMinLengthIndicator();
        MaxLengthIndicator maxLengthIndicator = IndicatorsFactory.eINSTANCE.createMaxLengthIndicator();
        AverageLengthIndicator averageLengthIndicator = IndicatorsFactory.eINSTANCE.createAverageLengthIndicator();
        AverageLengthIndicator averageLengthIndicator2 = IndicatorsFactory.eINSTANCE.createAverageLengthIndicator();
        ModeIndicator modeIndicator = IndicatorsFactory.eINSTANCE.createModeIndicator();
        FrequencyIndicator textFrequencyIndicator = IndicatorsFactory.eINSTANCE.createFrequencyIndicator();
        // store in freq indic
        // textFrequencyIndicator.setDistinctCountIndicator(distinctCountIndicator);
        // textFrequencyIndicator.setDistinctCountIndicator(distinctCountIndicator2);
        // textFrequencyIndicator.setUniqueCountIndicator(uniqueCountIndicator);
        // textFrequencyIndicator.setDuplicateCountIndicator(duplicateCountIndicator);
        // textFrequencyIndicator.setModeIndicator(modeIndicator);
        MeanIndicator doubleMeanIndicator = IndicatorsFactory.eINSTANCE.createMeanIndicator();
        MeanIndicator integerMeanIndicator = IndicatorsFactory.eINSTANCE.createMeanIndicator();
        MedianIndicator medianIndicator = IndicatorsFactory.eINSTANCE.createMedianIndicator();
        SumIndicator integerSumIndicator = IndicatorsFactory.eINSTANCE.createSumIndicator();
        addIndicator(columnsArray[0], medianIndicator);
        addIndicator(columnsArray[1], doubleMeanIndicator);
        addIndicator(columnsArray[2], blankCountIndicator);
        addIndicator(columnsArray[5], nullCountIndicator);
        // addIndicator(columnsArray[2], textFrequencyIndicator);
        // addIndicator(columnsArray[2], distinctCountIndicator); // probably not useful?
        // addIndicator(columnsArray[2], uniqueCountIndicator); // probably not useful?
        // addIndicator(columnsArray[2], duplicateCountIndicator); // probably not useful?
        // addIndicator(columnsArray[2], modeIndicator); // probably not useful?
        addIndicator(columnsArray[3], rowCountIndicator);
        addIndicator(columnsArray[5], integerSumIndicator);
        addIndicator(columnsArray[5], integerMeanIndicator);
        addIndicator(columnsArray[2], averageLengthIndicator);
        addIndicator(columnsArray[3], averageLengthIndicator2);
        addIndicator(columnsArray[3], minLengthIndicator);
        addIndicator(columnsArray[3], maxLengthIndicator);
        // build query on columns
        // TODO scorreia add filter somewhere here...
        String selectCols = sqlSelectColumns(database, tableName, columns);
        // --- create a description of the column set
        QueryExpression queryExpression = DatatypesFactory.eINSTANCE.createQueryExpression();
        queryExpression.setBody(selectCols);
        // TODO scorreia externalize this as a constant
        queryExpression.setLanguage("SQL");
        tt.start("compute");
        evaluator.setFetchSize(10000);
        evaluator.evaluateIndicators(selectCols, true);
        tt.end("compute");
        // Print indicators the median
        System.out.println("Median=" + medianIndicator.getMedian());
        System.out.println("# Unique values= " + textFrequencyIndicator.getUniqueValueCount());
        System.out.println("# Distinct values= " + textFrequencyIndicator.getDistinctValueCount());
        for (String col : columns) {
            printIndicators(evaluator.getIndicators(col));
        }
        tt.start("save");
        util.save();
        tt.end("saved in " + file.getAbsolutePath());
        tt.end();
        CwmResource cwmR = (CwmResource) resource;
        String id = cwmR.getID(medianIndicator);
        System.out.println("ecore util.getId= " + EcoreUtil.getID(medianIndicator));
        System.out.println("uuId= " + id);
    // test reload this file
    // LoadSerialData.main(args);
    } catch (SQLException e) {
        // TODO Auto-generated catch block
        log.error(e, e);
    } catch (InstantiationException e) {
        // TODO Auto-generated catch block
        log.error(e, e);
    } catch (IllegalAccessException e) {
        // TODO Auto-generated catch block
        log.error(e, e);
    } catch (ClassNotFoundException e) {
        // TODO Auto-generated catch block
        log.error(e, e);
    }
}
Also used : SumIndicator(org.talend.dataquality.indicators.SumIndicator) SQLException(java.sql.SQLException) BlankCountIndicator(org.talend.dataquality.indicators.BlankCountIndicator) MedianIndicator(org.talend.dataquality.indicators.MedianIndicator) UniqueCountIndicator(org.talend.dataquality.indicators.UniqueCountIndicator) ModeIndicator(org.talend.dataquality.indicators.ModeIndicator) NullCountIndicator(org.talend.dataquality.indicators.NullCountIndicator) FrequencyIndicator(org.talend.dataquality.indicators.FrequencyIndicator) DistinctCountIndicator(org.talend.dataquality.indicators.DistinctCountIndicator) RowCountIndicator(org.talend.dataquality.indicators.RowCountIndicator) QueryExpression(orgomg.cwm.foundation.datatypes.QueryExpression) MeanIndicator(org.talend.dataquality.indicators.MeanIndicator) CwmResource(org.talend.model.emf.CwmResource) DuplicateCountIndicator(org.talend.dataquality.indicators.DuplicateCountIndicator) MaxLengthIndicator(org.talend.dataquality.indicators.MaxLengthIndicator) Connection(java.sql.Connection) CwmResource(org.talend.model.emf.CwmResource) Resource(org.eclipse.emf.ecore.resource.Resource) MinLengthIndicator(org.talend.dataquality.indicators.MinLengthIndicator) TypedProperties(org.talend.utils.properties.TypedProperties) TimeTracer(org.talend.utils.time.TimeTracer) AverageLengthIndicator(org.talend.dataquality.indicators.AverageLengthIndicator) EMFUtil(org.talend.commons.emf.EMFUtil) File(java.io.File)

Aggregations

File (java.io.File)1 Connection (java.sql.Connection)1 SQLException (java.sql.SQLException)1 Resource (org.eclipse.emf.ecore.resource.Resource)1 EMFUtil (org.talend.commons.emf.EMFUtil)1 AverageLengthIndicator (org.talend.dataquality.indicators.AverageLengthIndicator)1 BlankCountIndicator (org.talend.dataquality.indicators.BlankCountIndicator)1 DistinctCountIndicator (org.talend.dataquality.indicators.DistinctCountIndicator)1 DuplicateCountIndicator (org.talend.dataquality.indicators.DuplicateCountIndicator)1 FrequencyIndicator (org.talend.dataquality.indicators.FrequencyIndicator)1 MaxLengthIndicator (org.talend.dataquality.indicators.MaxLengthIndicator)1 MeanIndicator (org.talend.dataquality.indicators.MeanIndicator)1 MedianIndicator (org.talend.dataquality.indicators.MedianIndicator)1 MinLengthIndicator (org.talend.dataquality.indicators.MinLengthIndicator)1 ModeIndicator (org.talend.dataquality.indicators.ModeIndicator)1 NullCountIndicator (org.talend.dataquality.indicators.NullCountIndicator)1 RowCountIndicator (org.talend.dataquality.indicators.RowCountIndicator)1 SumIndicator (org.talend.dataquality.indicators.SumIndicator)1 UniqueCountIndicator (org.talend.dataquality.indicators.UniqueCountIndicator)1 CwmResource (org.talend.model.emf.CwmResource)1