void Sessionize::describeDataflowAndPredicates(UDRInvocationInfo &info)
{
// Start with the default behavior for a reducer, pushing down
// any predicates on the key/id column.
UDR::describeDataflowAndPredicates(info);
// Make sure we don't require any unused passthru columns
// from the child/input table. NOTE: This can change the
// column numbers for our id and timestamp columns!
info.setUnusedPassthruColumns();
// That could have set our timestamp column or user id
// column as unused, however. So, make sure these two
// columns are definitely included.
// first, recompute the id and timestamp column numbers
InternalColumns state(
info.in().getColNum(info.par().getString(0)),
info.in().getColNum(info.par().getString(1)));
// then include the columns
info.setChildColumnUsage(0, state.getIdColumn(), ColumnInfo::USED);
info.setChildColumnUsage(0, state.getTsColumn(), ColumnInfo::USED);
bool generatedColsAreUsed =
(info.out().getColumn(0).getUsage() == ColumnInfo::USED ||
info.out().getColumn(1).getUsage() == ColumnInfo::USED);
// Walk through predicates and find additional ones to push down
// or to evaluate locally
for (int p=0; p<info.getNumPredicates(); p++)
{
if (!generatedColsAreUsed)
{
// If session_id/sequence_no are not used in the query, then
// we can push all predicates to the children.
info.setPredicateEvaluationCode(p, PredicateInfo::EVALUATE_IN_CHILD);
}
else if (info.isAComparisonPredicate(p))
{
// For demo purposes, accept predicates of the
// form "session_id < const" to be evaluated in the UDF.
const ComparisonPredicateInfo &cpi = info.getComparisonPredicate(p);
if (cpi.getColumnNumber() == 0 /* SESSION_ID */ &&
cpi.getOperator() == PredicateInfo::LESS &&
cpi.hasAConstantValue())
info.setPredicateEvaluationCode(p, PredicateInfo::EVALUATE_IN_UDF);
}
}
}
void Sessionize::describeStatistics(UDRInvocationInfo &info)
{
// We set the function type to REDUCER earlier. The Trafodion compiler
// estimates one output row per input partition for reducer function,
// unless the UDF specifies another value. Since our sessionize UDF
// returns one output row per input row, make sure the optimizer has
// a better cardinality estimate.
// Crude estimate, assume each predicate evaluated by the UDF
// reduces the number of output columns by 50%. At this point, only
// predicates that are evaluated by the UDF are left in the list.
double selectivity = pow(2,-info.getNumPredicates());
long resultRowCount =
static_cast<long>(info.in().getEstimatedNumRows() * selectivity);
info.out().setEstimatedNumRows(resultRowCount);
}
void Sessionize::processData(UDRInvocationInfo &info,
UDRPlanInfo &plan)
{
// read the three parameters and convert the first two into column numbers
int userIdColNum = info.in(0).getColNum(info.par().getString(0));
int timeStampColNum = info.in(0).getColNum(info.par().getString(1));
long timeout = info.par().getLong(2);
// variables needed for computing the session id
long lastTimeStamp = 0;
std::string lastUserId;
long currSessionId = 1;
long currSequenceNo = 1;
int maxSessionId = 999999999;
if (info.getNumPredicates() > 0)
{
// based on the describeDataflowAndPredicates() method, this must be
// a predicate of the form SESSION_ID < const that we need
// to evaluate inside this method
std::string maxValue = info.getComparisonPredicate(0).getConstValue();
sscanf(maxValue.c_str(), "%d", &maxSessionId);
}
// loop over input rows
while (getNextRow(info))
{
long timeStamp = info.in(0).getLong(timeStampColNum);
std::string userId = info.in(0).getString(userIdColNum);
if (lastUserId != userId)
{
// reset timestamp check and start over with session id 0
lastTimeStamp = 0;
currSessionId = 1;
currSequenceNo = 1;
lastUserId = userId;
}
long tsDiff = timeStamp - lastTimeStamp;
if (tsDiff > timeout && lastTimeStamp > 0)
{
currSessionId++;
currSequenceNo = 1;
}
else if (tsDiff < 0)
throw UDRException(
38001,
"Got negative or descending timestamps %ld, %ld",
lastTimeStamp, timeStamp);
lastTimeStamp = timeStamp;
// this evaluates the SQL predicate on SESSION_ID
if (currSessionId < maxSessionId)
{
// produce session_id and sequence_no output columns
info.out().setLong(0, currSessionId);
info.out().setLong(1, currSequenceNo);
// produce the remaining columns and emit the row
info.copyPassThruData();
emitRow(info);
currSequenceNo++;
}
}
}