Search in sources :

Example 16 with AnalyzedDataSet

use of org.talend.dataquality.analysis.AnalyzedDataSet in project tdq-studio-se by Talend.

the class IndicatorEvaluator method executeSqlQuery.

@Override
protected ReturnCode executeSqlQuery(String sqlStatement) throws SQLException {
    ReturnCode ok = new ReturnCode(true);
    // check analyzed columns
    Set<String> columns = getAnalyzedElements();
    // feature 0010630 zshen:Make order unify which columns and columnName in the sqlStatement.mssqlOdbc need do
    // this
    List<String> columnlist = sortColumnName(columns, sqlStatement);
    if (columnlist.isEmpty()) {
        // $NON-NLS-1$
        ok.setReturnCode(Messages.getString("IndicatorEvaluator.DefineAnalyzedColumns"), false);
        return ok;
    }
    // ADD xqliu 2010-07-27 bug 13826
    Map<String, String> columnlistMap = buildColumnListMap(columnlist);
    // ~ 13826
    // create query statement
    // feature 0010630 zshen: Tables are not found when using Excel with ODBC connection
    Statement statement = createStatement();
    // MOD xqliu 2009-02-09 bug 6237
    if (continueRun()) {
        if (log.isInfoEnabled()) {
            // $NON-NLS-1$
            log.info("Executing query: " + sqlStatement);
        }
        statement.execute(sqlStatement);
    }
    // get the results
    ResultSet resultSet = statement.getResultSet();
    if (resultSet == null) {
        // $NON-NLS-1$
        String mess = Messages.getString("Evaluator.NoResultSet", sqlStatement);
        log.warn(mess);
        ok.setReturnCode(mess, false);
        statement.close();
        return ok;
    }
    // MOD qiongli TDQ-7282,check invalid judi.if there are invalid judis,return false code and show message later.
    ok = getMessageForInvalidJUDIs();
    int columnCount = resultSet.getMetaData().getColumnCount();
    int maxNumberRows = analysis.getParameters().getMaxNumberRows();
    // MOD mzhao feature: 12919, add capability to dill down data on Java engine.
    AnalysisResult anaResult = analysis.getResults();
    EMap<Indicator, AnalyzedDataSet> indicToRowMap = anaResult.getIndicToRowMap();
    indicToRowMap.clear();
    int recordIncrement = 0;
    // --- for each row
    int columnListSize = columnlist.size();
    label: while (resultSet.next()) {
        // feature 0010630 zshen: dislodge the Qualifiers from name of the column
        for (int i = 0; i < columnListSize; i++) {
            // MOD xqliu 2010-07-27 bug 13826
            String col = columnlist.get(i);
            List<Indicator> indicators = getIndicators(col);
            col = columnlistMap.get(col);
            // --- get content of column
            Object object = ResultSetUtils.getBigObject(resultSet, col);
            // FIXME this will slow down a lot the computation
            if (object != null && !(object instanceof String) && object.toString().indexOf("TIMESTAMP") > -1) {
                // $NON-NLS-1$
                object = resultSet.getTimestamp(col);
            }
            // TDQ-11299: fix the ClassCastException: java.sql.Date cannot be cast to java.lang.String
            if (object instanceof Date) {
                if (object instanceof Time) {
                    object = new TalendFormatTime((Time) object);
                } else {
                    object = new TalendFormatDate((Date) object);
                }
            }
            // --- give row to handle to indicators
            for (Indicator indicator : indicators) {
                // MOD xqliu 2009-02-09 bug 6237
                if (!continueRun()) {
                    break label;
                }
                // Added yyin 20120608 TDQ-3589
                if (indicator instanceof DuplicateCountIndicator) {
                    ((DuplicateCountIndicator) indicator).handle(object, resultSet, columnCount);
                } else {
                    // ~
                    indicator.handle(object);
                // ~MOD mzhao feature: 12919
                }
                AnalyzedDataSet analyzedDataSet = indicToRowMap.get(indicator);
                if (analyzedDataSet == null) {
                    analyzedDataSet = AnalysisFactory.eINSTANCE.createAnalyzedDataSet();
                    indicToRowMap.put(indicator, analyzedDataSet);
                    analyzedDataSet.setDataCount(maxNumberRows);
                    analyzedDataSet.setRecordSize(0);
                }
                // should store data for dirll down
                if (analysis.getParameters().isStoreData()) {
                    // current indicator is need to store the data
                    if (indicator.mustStoreRow()) {
                        List<Object[]> valueObjectList = initDataSet(indicator, indicToRowMap, object);
                        // MOD zshen add another loop to insert all of columnValue on the row into indicator.
                        recordIncrement = valueObjectList.size();
                        // MOD klliu 2011-06-30 bug 22523 whichever is Table or View,that finds columns should ues
                        // columnset
                        ColumnSet doSwitch = SwitchHelpers.COLUMN_SET_SWITCH.doSwitch(indicator.getAnalyzedElement().eContainer());
                        List<TdColumn> columnList = ColumnSetHelper.getColumns(doSwitch);
                        List<Object> inputRowList = new ArrayList<Object>();
                        for (int j = 0; j < columnCount; j++) {
                            String newcol = columnList.get(j).getName();
                            Object newobject = ResultSetUtils.getBigObject(resultSet, newcol);
                            // same format as result page.
                            if (newobject instanceof Date) {
                                if (newobject instanceof Time) {
                                    newobject = new TalendFormatTime((Time) newobject);
                                } else {
                                    newobject = new TalendFormatDate((Date) newobject);
                                }
                            }
                            if (indicator.isUsedMapDBMode()) {
                                inputRowList.add(newobject == null ? PluginConstant.NULL_STRING : newobject);
                                continue;
                            } else {
                                if (recordIncrement < maxNumberRows) {
                                    // data.
                                    if (recordIncrement < valueObjectList.size()) {
                                        // decide whether need to
                                        // increase
                                        // current array.
                                        valueObjectList.get(recordIncrement)[j] = newobject;
                                    } else {
                                        Object[] valueObject = new Object[columnCount];
                                        valueObject[j] = newobject;
                                        valueObjectList.add(valueObject);
                                    }
                                } else {
                                    break;
                                }
                            }
                        }
                        if (indicator.isUsedMapDBMode()) {
                            MapDBUtils.handleDrillDownData(object, inputRowList, indicator);
                        }
                    // ~
                    } else if (indicator instanceof UniqueCountIndicator && analysis.getResults().getIndicToRowMap().get(indicator).getData() != null) {
                        List<Object[]> removeValueObjectList = analysis.getResults().getIndicToRowMap().get(indicator).getData();
                        // MOD klliu 2011-06-30 bug 22523 whichever is Table or View,that finds columns should ues
                        // columnset
                        ColumnSet doSwitch = SwitchHelpers.COLUMN_SET_SWITCH.doSwitch(indicator.getAnalyzedElement().eContainer());
                        List<TdColumn> columnElementList = ColumnSetHelper.getColumns(doSwitch);
                        int offsetting = columnElementList.indexOf(indicator.getAnalyzedElement());
                        for (Object[] dataObject : removeValueObjectList) {
                            if (dataObject[offsetting].equals(object)) {
                                removeValueObjectList.remove(dataObject);
                                break;
                            }
                        }
                    }
                }
            }
        }
    }
    // --- release resultset
    resultSet.close();
    // --- release statement
    statement.close();
    // --- close connection
    getConnection().close();
    return ok;
}
Also used : UniqueCountIndicator(org.talend.dataquality.indicators.UniqueCountIndicator) Time(java.sql.Time) TalendFormatTime(org.talend.dataquality.indicators.mapdb.TalendFormatTime) ColumnSet(orgomg.cwm.resource.relational.ColumnSet) TalendFormatTime(org.talend.dataquality.indicators.mapdb.TalendFormatTime) ResultSet(java.sql.ResultSet) ArrayList(java.util.ArrayList) List(java.util.List) DuplicateCountIndicator(org.talend.dataquality.indicators.DuplicateCountIndicator) ReturnCode(org.talend.utils.sugars.ReturnCode) AnalyzedDataSet(org.talend.dataquality.analysis.AnalyzedDataSet) TalendFormatDate(org.talend.dataquality.indicators.mapdb.TalendFormatDate) Statement(java.sql.Statement) AnalysisResult(org.talend.dataquality.analysis.AnalysisResult) UniqueCountIndicator(org.talend.dataquality.indicators.UniqueCountIndicator) FrequencyIndicator(org.talend.dataquality.indicators.FrequencyIndicator) FormatFreqPieIndicator(org.talend.dataquality.indicators.FormatFreqPieIndicator) MaxLengthIndicator(org.talend.dataquality.indicators.MaxLengthIndicator) Indicator(org.talend.dataquality.indicators.Indicator) MinLengthIndicator(org.talend.dataquality.indicators.MinLengthIndicator) PatternLowFreqIndicator(org.talend.dataquality.indicators.PatternLowFreqIndicator) PatternFreqIndicator(org.talend.dataquality.indicators.PatternFreqIndicator) DuplicateCountIndicator(org.talend.dataquality.indicators.DuplicateCountIndicator) UserDefIndicator(org.talend.dataquality.indicators.sql.UserDefIndicator) Date(java.util.Date) TalendFormatDate(org.talend.dataquality.indicators.mapdb.TalendFormatDate) TdColumn(org.talend.cwm.relational.TdColumn)

Aggregations

AnalyzedDataSet (org.talend.dataquality.analysis.AnalyzedDataSet)16 Indicator (org.talend.dataquality.indicators.Indicator)11 DuplicateCountIndicator (org.talend.dataquality.indicators.DuplicateCountIndicator)8 UniqueCountIndicator (org.talend.dataquality.indicators.UniqueCountIndicator)8 RowCountIndicator (org.talend.dataquality.indicators.RowCountIndicator)7 ArrayList (java.util.ArrayList)6 DistinctCountIndicator (org.talend.dataquality.indicators.DistinctCountIndicator)5 SimpleStatIndicator (org.talend.dataquality.indicators.columnset.SimpleStatIndicator)5 ReturnCode (org.talend.utils.sugars.ReturnCode)5 File (java.io.File)4 List (java.util.List)4 IPath (org.eclipse.core.runtime.IPath)4 MetadataColumn (org.talend.core.model.metadata.builder.connection.MetadataColumn)4 AnalysisResult (org.talend.dataquality.analysis.AnalysisResult)4 AllMatchIndicator (org.talend.dataquality.indicators.columnset.AllMatchIndicator)4 ModelElement (orgomg.cwm.objectmodel.core.ModelElement)4 BasicEList (org.eclipse.emf.common.util.BasicEList)3 DelimitedFileConnection (org.talend.core.model.metadata.builder.connection.DelimitedFileConnection)3 MetadataTable (org.talend.core.model.metadata.builder.connection.MetadataTable)3 FrequencyIndicator (org.talend.dataquality.indicators.FrequencyIndicator)3