use of org.pentaho.di.core.row.value.ValueMetaInteger in project pentaho-kettle by pentaho.
the class GroupBy method processRow.
@Override
public boolean processRow(StepMetaInterface smi, StepDataInterface sdi) throws KettleException {
meta = (GroupByMeta) smi;
data = (GroupByData) sdi;
// get row!
Object[] r = getRow();
if (first) {
String val = getVariable(Const.KETTLE_AGGREGATION_ALL_NULLS_ARE_ZERO, "N");
allNullsAreZero = ValueMetaBase.convertStringToBoolean(val);
val = getVariable(Const.KETTLE_AGGREGATION_MIN_NULL_IS_VALUED, "N");
minNullIsValued = ValueMetaBase.convertStringToBoolean(val);
// What is the output looking like?
//
data.inputRowMeta = getInputRowMeta();
//
if (data.inputRowMeta == null) {
data.inputRowMeta = getTransMeta().getPrevStepFields(getStepMeta());
}
data.outputRowMeta = data.inputRowMeta.clone();
meta.getFields(data.outputRowMeta, getStepname(), null, null, this, repository, metaStore);
// Do all the work we can beforehand
// Calculate indexes, loop up fields, etc.
//
data.counts = new long[meta.getSubjectField().length];
data.subjectnrs = new int[meta.getSubjectField().length];
data.cumulativeSumSourceIndexes = new ArrayList<>();
data.cumulativeSumTargetIndexes = new ArrayList<>();
data.cumulativeAvgSourceIndexes = new ArrayList<>();
data.cumulativeAvgTargetIndexes = new ArrayList<>();
for (int i = 0; i < meta.getSubjectField().length; i++) {
if (meta.getAggregateType()[i] == GroupByMeta.TYPE_GROUP_COUNT_ANY) {
data.subjectnrs[i] = 0;
} else {
data.subjectnrs[i] = data.inputRowMeta.indexOfValue(meta.getSubjectField()[i]);
}
if ((r != null) && (data.subjectnrs[i] < 0)) {
logError(BaseMessages.getString(PKG, "GroupBy.Log.AggregateSubjectFieldCouldNotFound", meta.getSubjectField()[i]));
setErrors(1);
stopAll();
return false;
}
if (meta.getAggregateType()[i] == GroupByMeta.TYPE_GROUP_CUMULATIVE_SUM) {
data.cumulativeSumSourceIndexes.add(data.subjectnrs[i]);
// The position of the target in the output row is the input row size + i
//
data.cumulativeSumTargetIndexes.add(data.inputRowMeta.size() + i);
}
if (meta.getAggregateType()[i] == GroupByMeta.TYPE_GROUP_CUMULATIVE_AVERAGE) {
data.cumulativeAvgSourceIndexes.add(data.subjectnrs[i]);
// The position of the target in the output row is the input row size + i
//
data.cumulativeAvgTargetIndexes.add(data.inputRowMeta.size() + i);
}
}
data.previousSums = new Object[data.cumulativeSumTargetIndexes.size()];
data.previousAvgSum = new Object[data.cumulativeAvgTargetIndexes.size()];
data.previousAvgCount = new long[data.cumulativeAvgTargetIndexes.size()];
data.groupnrs = new int[meta.getGroupField().length];
for (int i = 0; i < meta.getGroupField().length; i++) {
data.groupnrs[i] = data.inputRowMeta.indexOfValue(meta.getGroupField()[i]);
if ((r != null) && (data.groupnrs[i] < 0)) {
logError(BaseMessages.getString(PKG, "GroupBy.Log.GroupFieldCouldNotFound", meta.getGroupField()[i]));
setErrors(1);
stopAll();
return false;
}
}
// Create a metadata value for the counter Integers
//
data.valueMetaInteger = new ValueMetaInteger("count");
data.valueMetaNumber = new ValueMetaNumber("sum");
// Initialize the group metadata
//
initGroupMeta(data.inputRowMeta);
}
if (first || data.newBatch) {
// Create a new group aggregate (init)
//
newAggregate(r);
}
if (first) {
// for speed: groupMeta+aggMeta
//
data.groupAggMeta = new RowMeta();
data.groupAggMeta.addRowMeta(data.groupMeta);
data.groupAggMeta.addRowMeta(data.aggMeta);
}
if (r == null) {
// no more input to be expected... (or none received in the first place)
handleLastOfGroup();
setOutputDone();
return false;
}
if (first || data.newBatch) {
first = false;
data.newBatch = false;
// copy the row to previous
data.previous = data.inputRowMeta.cloneRow(r);
} else {
calcAggregate(data.previous);
if (meta.passAllRows()) {
addToBuffer(data.previous);
}
}
if (!sameGroup(data.previous, r)) {
if (meta.passAllRows()) {
// Not the same group: close output (if any)
closeOutput();
// Get all rows from the buffer!
data.groupResult = getAggregateResult();
Object[] row = getRowFromBuffer();
long lineNr = 0;
while (row != null) {
int size = data.inputRowMeta.size();
row = RowDataUtil.addRowData(row, size, data.groupResult);
size += data.groupResult.length;
lineNr++;
if (meta.isAddingLineNrInGroup() && !Utils.isEmpty(meta.getLineNrInGroupField())) {
Object lineNrValue = new Long(lineNr);
// ValueMetaInterface lineNrValueMeta = new ValueMeta(meta.getLineNrInGroupField(),
// ValueMetaInterface.TYPE_INTEGER);
// lineNrValueMeta.setLength(9);
row = RowDataUtil.addValueData(row, size, lineNrValue);
size++;
}
addCumulativeSums(row);
addCumulativeAverages(row);
putRow(data.outputRowMeta, row);
row = getRowFromBuffer();
}
closeInput();
} else {
Object[] result = buildResult(data.previous);
if (result != null) {
// copy row to possible alternate rowset(s).
putRow(data.groupAggMeta, result);
}
}
// Create a new group aggregate (init)
newAggregate(r);
}
data.previous = data.inputRowMeta.cloneRow(r);
if (checkFeedback(getLinesRead())) {
if (log.isBasic()) {
logBasic(BaseMessages.getString(PKG, "GroupBy.LineNumber") + getLinesRead());
}
}
return true;
}
use of org.pentaho.di.core.row.value.ValueMetaInteger in project pentaho-kettle by pentaho.
the class GroupBy method newAggregate.
/**
* used for junits in GroupByAggregationNullsTest
*
* @param r
*/
void newAggregate(Object[] r) {
// Put all the counters at 0
for (int i = 0; i < data.counts.length; i++) {
data.counts[i] = 0;
}
data.distinctObjs = null;
data.agg = new Object[data.subjectnrs.length];
// sets all doubles to 0.0
data.mean = new double[data.subjectnrs.length];
data.aggMeta = new RowMeta();
for (int i = 0; i < data.subjectnrs.length; i++) {
ValueMetaInterface subjMeta = data.inputRowMeta.getValueMeta(data.subjectnrs[i]);
Object v = null;
ValueMetaInterface vMeta = null;
int aggType = meta.getAggregateType()[i];
switch(aggType) {
case GroupByMeta.TYPE_GROUP_SUM:
case GroupByMeta.TYPE_GROUP_AVERAGE:
case GroupByMeta.TYPE_GROUP_CUMULATIVE_SUM:
case GroupByMeta.TYPE_GROUP_CUMULATIVE_AVERAGE:
if (subjMeta.isNumeric()) {
try {
vMeta = ValueMetaFactory.createValueMeta(meta.getAggregateField()[i], subjMeta.getType());
} catch (KettlePluginException e) {
vMeta = new ValueMetaNone(meta.getAggregateField()[i]);
}
} else {
vMeta = new ValueMetaNumber(meta.getAggregateField()[i]);
}
break;
case GroupByMeta.TYPE_GROUP_MEDIAN:
case GroupByMeta.TYPE_GROUP_PERCENTILE:
case GroupByMeta.TYPE_GROUP_PERCENTILE_NEAREST_RANK:
vMeta = new ValueMetaNumber(meta.getAggregateField()[i]);
v = new ArrayList<Double>();
break;
case GroupByMeta.TYPE_GROUP_STANDARD_DEVIATION:
case GroupByMeta.TYPE_GROUP_STANDARD_DEVIATION_SAMPLE:
vMeta = new ValueMetaNumber(meta.getAggregateField()[i]);
break;
case GroupByMeta.TYPE_GROUP_COUNT_DISTINCT:
case GroupByMeta.TYPE_GROUP_COUNT_ANY:
case GroupByMeta.TYPE_GROUP_COUNT_ALL:
vMeta = new ValueMetaInteger(meta.getAggregateField()[i]);
break;
case GroupByMeta.TYPE_GROUP_FIRST:
case GroupByMeta.TYPE_GROUP_LAST:
case GroupByMeta.TYPE_GROUP_FIRST_INCL_NULL:
case GroupByMeta.TYPE_GROUP_LAST_INCL_NULL:
case GroupByMeta.TYPE_GROUP_MIN:
case GroupByMeta.TYPE_GROUP_MAX:
vMeta = subjMeta.clone();
vMeta.setName(meta.getAggregateField()[i]);
v = r == null ? null : r[data.subjectnrs[i]];
break;
case GroupByMeta.TYPE_GROUP_CONCAT_COMMA:
vMeta = new ValueMetaString(meta.getAggregateField()[i]);
v = new StringBuilder();
break;
case GroupByMeta.TYPE_GROUP_CONCAT_STRING:
vMeta = new ValueMetaString(meta.getAggregateField()[i]);
v = new StringBuilder();
break;
default:
// TODO raise an error here because we cannot continue successfully maybe the UI should validate this
break;
}
if ((subjMeta != null) && (aggType != GroupByMeta.TYPE_GROUP_COUNT_ALL && aggType != GroupByMeta.TYPE_GROUP_COUNT_DISTINCT && aggType != GroupByMeta.TYPE_GROUP_COUNT_ANY)) {
vMeta.setLength(subjMeta.getLength(), subjMeta.getPrecision());
}
data.agg[i] = v;
data.aggMeta.addValueMeta(vMeta);
}
//
for (int i = 0; i < data.previousSums.length; i++) {
data.previousSums[i] = null;
}
for (int i = 0; i < data.previousAvgCount.length; i++) {
data.previousAvgCount[i] = 0L;
data.previousAvgSum[i] = null;
}
}
use of org.pentaho.di.core.row.value.ValueMetaInteger in project pentaho-kettle by pentaho.
the class BaseMetadataInjectionTest method check.
/**
* Check int property.
*/
protected void check(String propertyName, IntGetter getter) throws KettleException {
ValueMetaInterface valueMetaString = new ValueMetaString("f");
injector.setProperty(meta, propertyName, setValue(valueMetaString, "1"), "f");
assertEquals(1, getter.get());
injector.setProperty(meta, propertyName, setValue(valueMetaString, "45"), "f");
assertEquals(45, getter.get());
ValueMetaInterface valueMetaInteger = new ValueMetaInteger("f");
injector.setProperty(meta, propertyName, setValue(valueMetaInteger, 1234L), "f");
assertEquals(1234, getter.get());
injector.setProperty(meta, propertyName, setValue(valueMetaInteger, (long) Integer.MAX_VALUE), "f");
assertEquals(Integer.MAX_VALUE, getter.get());
skipPropertyTest(propertyName);
}
use of org.pentaho.di.core.row.value.ValueMetaInteger in project pentaho-kettle by pentaho.
the class BaseMetadataInjectionTest method check.
/**
* Check long property.
*/
protected void check(String propertyName, LongGetter getter) throws KettleException {
ValueMetaInterface valueMetaString = new ValueMetaString("f");
injector.setProperty(meta, propertyName, setValue(valueMetaString, "1"), "f");
assertEquals(1, getter.get());
injector.setProperty(meta, propertyName, setValue(valueMetaString, "45"), "f");
assertEquals(45, getter.get());
ValueMetaInterface valueMetaInteger = new ValueMetaInteger("f");
injector.setProperty(meta, propertyName, setValue(valueMetaInteger, 1234L), "f");
assertEquals(1234, getter.get());
injector.setProperty(meta, propertyName, setValue(valueMetaInteger, Long.MAX_VALUE), "f");
assertEquals(Long.MAX_VALUE, getter.get());
skipPropertyTest(propertyName);
}
use of org.pentaho.di.core.row.value.ValueMetaInteger in project pentaho-kettle by pentaho.
the class GroupByMetaGetFieldsTest method setup.
@Before
public void setup() throws KettlePluginException {
rowMeta = spy(new RowMeta());
groupByMeta = spy(new GroupByMeta());
mockStatic(ValueMetaFactory.class);
when(ValueMetaFactory.createValueMeta(anyInt())).thenCallRealMethod();
when(ValueMetaFactory.createValueMeta(anyString(), anyInt())).thenCallRealMethod();
when(ValueMetaFactory.createValueMeta("maxDate", 3, -1, -1)).thenReturn(new ValueMetaDate("maxDate"));
when(ValueMetaFactory.createValueMeta("minDate", 3, -1, -1)).thenReturn(new ValueMetaDate("minDate"));
when(ValueMetaFactory.createValueMeta("countDate", 5, -1, -1)).thenReturn(new ValueMetaInteger("countDate"));
when(ValueMetaFactory.getValueMetaName(3)).thenReturn("Date");
when(ValueMetaFactory.getValueMetaName(5)).thenReturn("Integer");
}
Aggregations