/**
* Initialize the step.
*
* @param smi a <code>StepMetaInterface</code> value
* @param sdi a <code>StepDataInterface</code> value
* @return a <code>boolean</code> value
*/
public boolean init(StepMetaInterface smi, StepDataInterface sdi) {
m_meta = (ReservoirSamplingMeta) smi;
m_data = (ReservoirSamplingData) sdi;
if (super.init(smi, sdi)) {
List<StepMeta> previous = getTransMeta().findPreviousSteps(getStepMeta());
if (previous == null || previous.size() <= 0) {
m_data.setProcessingMode(PROC_MODE.DISABLED);
}
return true;
}
return false;
}
/**
* Process an incoming row of data.
*
* @param smi a <code>StepMetaInterface</code> value
* @param sdi a <code>StepDataInterface</code> value
* @return a <code>boolean</code> value
* @exception KettleException if an error occurs
*/
public boolean processRow(StepMetaInterface smi, StepDataInterface sdi) throws KettleException {
if (m_data.getProcessingMode() == PROC_MODE.DISABLED) {
setOutputDone();
m_data.cleanUp();
return (false);
}
m_meta = (ReservoirSamplingMeta) smi;
m_data = (ReservoirSamplingData) sdi;
Object[] r = getRow();
// Handle the first row
if (first) {
first = false;
if (r == null) // no input to be expected...
{
setOutputDone();
return false;
}
// Initialize the data object
m_data.setOutputRowMeta(getInputRowMeta().clone());
String sampleSize = getTransMeta().environmentSubstitute(m_meta.getSampleSize());
String seed = getTransMeta().environmentSubstitute(m_meta.getSeed());
m_data.initialize(Integer.valueOf(sampleSize), Integer.valueOf(seed));
// no real reason to determine the output fields here
// as we don't add/delete any fields
} // end (if first)
if (m_data.getProcessingMode() == PROC_MODE.PASSTHROUGH) {
if (r == null) {
setOutputDone();
m_data.cleanUp();
return (false);
}
putRow(m_data.getOutputRowMeta(), r);
} else if (m_data.getProcessingMode() == PROC_MODE.SAMPLING) {
if (r == null) {
// Output the rows in the sample
List<Object[]> samples = m_data.getSample();
int numRows = (samples != null) ? samples.size() : 0;
logBasic(
this.getStepname()
+ " Actual/Sample: "
+ numRows
+ "/"
+ m_data.m_k
+ " Seed:"
+ getTransMeta().environmentSubstitute(m_meta.m_randomSeed));
if (samples != null) {
for (int i = 0; i < samples.size(); i++) {
Object[] sample = samples.get(i);
if (sample != null) {
putRow(m_data.getOutputRowMeta(), sample);
} else {
// user probably requested more rows in
// the sample than there were in total
// in the end. Just break in this case
break;
}
}
}
setOutputDone();
m_data.cleanUp();
return false;
}
// just pass the row to the data class for possible caching
// in the sample
m_data.processRow(r);
}
if (log.isRowLevel()) {
log.logRowlevel(toString(), "Read row #" + getLinesRead() + " : " + r);
}
if (checkFeedback(getLinesRead())) {
logBasic("Line number " + getLinesRead());
}
return true;
}
