Examples with DataColumnSpec org.knime.core.data.DataColumnSpec used on opensource projects

Example 91 with DataColumnSpec

use of org.knime.core.data.DataColumnSpec in project knime-core by knime.

the class NormalizerNodeModel method calculate.

/**
 * New normalized {@link org.knime.core.data.DataTable} is created depending
 * on the mode.
 */
/**
 * @param inData The input data.
 * @param exec For BufferedDataTable creation and progress.
 * @return the result of the calculation
 * @throws Exception If the node calculation fails for any reason.
 */
protected CalculationResult calculate(final PortObject[] inData, final ExecutionContext exec) throws Exception {
    BufferedDataTable inTable = (BufferedDataTable) inData[0];
    DataTableSpec inSpec = inTable.getSpec();
    // extract selected numeric columns
    updateNumericColumnSelection(inSpec);
    Normalizer ntable = new Normalizer(inTable, m_columns);
    long rowcount = inTable.size();
    ExecutionMonitor prepareExec = exec.createSubProgress(0.3);
    AffineTransTable outTable;
    boolean fixDomainBounds = false;
    switch(m_mode) {
        case NONORM_MODE:
            return new CalculationResult(inTable, new DataTableSpec(), new AffineTransConfiguration());
        case MINMAX_MODE:
            fixDomainBounds = true;
            outTable = ntable.doMinMaxNorm(m_max, m_min, prepareExec);
            break;
        case ZSCORE_MODE:
            outTable = ntable.doZScoreNorm(prepareExec);
            break;
        case DECIMALSCALING_MODE:
            outTable = ntable.doDecimalScaling(prepareExec);
            break;
        default:
            throw new Exception("No mode set");
    }
    if (outTable.getErrorMessage() != null) {
        // something went wrong, report and throw an exception
        throw new Exception(outTable.getErrorMessage());
    }
    if (ntable.getErrorMessage() != null) {
        // something went wrong during initialization, report.
        setWarningMessage(ntable.getErrorMessage());
    }
    DataTableSpec modelSpec = FilterColumnTable.createFilterTableSpec(inSpec, m_columns);
    AffineTransConfiguration configuration = outTable.getConfiguration();
    DataTableSpec spec = outTable.getDataTableSpec();
    // the same transformation, which is not guaranteed to snap to min/max)
    if (fixDomainBounds) {
        DataColumnSpec[] newColSpecs = new DataColumnSpec[spec.getNumColumns()];
        for (int i = 0; i < newColSpecs.length; i++) {
            newColSpecs[i] = spec.getColumnSpec(i);
        }
        for (int i = 0; i < m_columns.length; i++) {
            int index = spec.findColumnIndex(m_columns[i]);
            DataColumnSpecCreator creator = new DataColumnSpecCreator(newColSpecs[index]);
            DataColumnDomainCreator domCreator = new DataColumnDomainCreator(newColSpecs[index].getDomain());
            domCreator.setLowerBound(new DoubleCell(m_min));
            domCreator.setUpperBound(new DoubleCell(m_max));
            creator.setDomain(domCreator.createDomain());
            newColSpecs[index] = creator.createSpec();
        }
        spec = new DataTableSpec(spec.getName(), newColSpecs);
    }
    ExecutionMonitor normExec = exec.createSubProgress(.7);
    BufferedDataContainer container = exec.createDataContainer(spec);
    long count = 1;
    for (DataRow row : outTable) {
        normExec.checkCanceled();
        normExec.setProgress(count / (double) rowcount, "Normalizing row no. " + count + " of " + rowcount + " (\"" + row.getKey() + "\")");
        container.addRowToTable(row);
        count++;
    }
    container.close();
    return new CalculationResult(container.getTable(), modelSpec, configuration);
}

Example 92 with DataColumnSpec

use of org.knime.core.data.DataColumnSpec in project knime-core by knime.

the class CollectionSplitNodeModel method getTargetColIndex.

/**
 * Validate settings and get the target column index.
 */
private int getTargetColIndex(final DataTableSpec spec) throws InvalidSettingsException {
    String colName = m_settings.getCollectionColName();
    if (colName == null || colName.length() == 0) {
        throw new InvalidSettingsException("Not configured");
    }
    final int colIndex = spec.findColumnIndex(colName);
    if (colIndex < 0) {
        throw new InvalidSettingsException("No such column: " + colName);
    }
    DataColumnSpec cs = spec.getColumnSpec(colIndex);
    if (!cs.getType().isCompatible(CollectionDataValue.class)) {
        throw new InvalidSettingsException("Column \"" + colName + "\" does not contain collection.");
    }
    return colIndex;
}

Example 93 with DataColumnSpec

use of org.knime.core.data.DataColumnSpec in project knime-core by knime.

the class CollectionSplitNodeModel method getColSpecsByElementNames.

/**
 * Get new column specs as inferred from the element names in the
 * collection column.
 */
private DataColumnSpec[] getColSpecsByElementNames(final DataTableSpec spec) throws InvalidSettingsException {
    int colIndex = getTargetColIndex(spec);
    DataColumnSpec colSpec = spec.getColumnSpec(colIndex);
    List<String> elementNames = colSpec.getElementNames();
    if (elementNames.isEmpty()) {
        throw new InvalidSettingsException("Input column \"" + colSpec.getName() + "\" does not provide element names; " + "consider to change option in dialog or make sure that" + "the input table contains the necessary information.");
    }
    DataType type = colSpec.getType().getCollectionElementType();
    HashSet<String> hashNames = new HashSet<String>();
    for (DataColumnSpec s : spec) {
        hashNames.add(s.getName());
    }
    if (m_settings.isReplaceInputColumn()) {
        hashNames.remove(colSpec.getName());
    }
    DataColumnSpec[] newColSpec = new DataColumnSpec[elementNames.size()];
    for (int i = 0; i < newColSpec.length; i++) {
        String baseName = elementNames.get(i);
        int uniquifier = 1;
        while (!hashNames.add(baseName)) {
            baseName = elementNames.get(i) + "(#" + (uniquifier++) + ")";
        }
        newColSpec[i] = new DataColumnSpecCreator(baseName, type).createSpec();
    }
    return newColSpec;
}

Example 94 with DataColumnSpec

use of org.knime.core.data.DataColumnSpec in project knime-core by knime.

the class CollectionSplitNodeModel method execute.

/**
 * {@inheritDoc}
 */
@Override
protected BufferedDataTable[] execute(final BufferedDataTable[] inData, final ExecutionContext exec) throws Exception {
    BufferedDataTable table = inData[0];
    DataTableSpec spec = table.getDataTableSpec();
    ExecutionMonitor execForCR = exec;
    // validate settings
    getTargetColIndex(spec);
    DataColumnSpec[] colSpecs;
    switch(m_settings.getCountElementsPolicy()) {
        case Count:
            execForCR = exec.createSubProgress(0.7);
            ExecutionMonitor e = exec.createSubProgress(0.3);
            colSpecs = countNewColumns(table, e);
            break;
        case UseElementNamesOrFail:
            colSpecs = getColSpecsByElementNames(spec);
            break;
        case BestEffort:
            try {
                colSpecs = getColSpecsByElementNames(spec);
            } catch (InvalidSettingsException ise) {
                execForCR = exec.createSubProgress(0.7);
                e = exec.createSubProgress(0.3);
                colSpecs = countNewColumns(table, e);
            }
            break;
        default:
            throw new InvalidSettingsException("Unsupported policy: " + m_settings.getCountElementsPolicy());
    }
    Pair<ColumnRearranger, SplitCellFactory> pair = createColumnRearranger(spec, colSpecs);
    BufferedDataTable out = exec.createColumnRearrangeTable(table, pair.getFirst(), execForCR);
    String warnMessage = pair.getSecond().getWarnMessage();
    if (warnMessage != null) {
        setWarningMessage(warnMessage);
    }
    if (m_settings.isDetermineMostSpecificDataType()) {
        out = refineTypes(out, pair.getSecond(), exec);
    }
    return new BufferedDataTable[] { out };
}

Example 95 with DataColumnSpec

use of org.knime.core.data.DataColumnSpec in project knime-core by knime.

the class CollectionSplitNodeModel method refineTypes.

/**
 * Retype the argument table to use the types as determined by the
 * cell factory.
 */
private BufferedDataTable refineTypes(final BufferedDataTable table, final SplitCellFactory fac, final ExecutionContext exec) {
    HashMap<String, Integer> colMap = new HashMap<String, Integer>();
    DataTableSpec spec = table.getDataTableSpec();
    DataColumnSpec[] newColSpecs = new DataColumnSpec[spec.getNumColumns()];
    for (int i = 0; i < spec.getNumColumns(); i++) {
        colMap.put(spec.getColumnSpec(i).getName(), i);
        newColSpecs[i] = spec.getColumnSpec(i);
    }
    DataColumnSpec[] oldReplacedSpecs = fac.getColumnSpecs();
    DataType[] mostSpecificTypes = fac.getCommonTypes();
    DataColumnDomain[] domains = fac.getDomains();
    for (int i = 0; i < oldReplacedSpecs.length; i++) {
        DataColumnSpec s = oldReplacedSpecs[i];
        Integer index = colMap.get(s.getName());
        DataColumnSpecCreator creator = new DataColumnSpecCreator(newColSpecs[index]);
        creator.setType(mostSpecificTypes[i]);
        creator.setDomain(domains[i]);
        newColSpecs[index] = creator.createSpec();
    }
    DataTableSpec newSpec = new DataTableSpec(spec.getName(), newColSpecs);
    return exec.createSpecReplacerTable(table, newSpec);
}