LbmRepeatingAggComputer Class Reference

LbmRepeatingAggComputer is an aggregate computer that wraps another aggregate computer. More...

Inheritance diagram for LbmRepeatingAggComputer:

AggComputer List of all members.

Public Member Functions

 LbmRepeatingAggComputer (AggComputer *pComputer)
virtual void setInputAttrIndex (uint iInputAttrIndex)
 Sets the attribute index from which this computer should read input values in source tuples.
virtual void clearAccumulator (TupleDatum &accumulatorDatum)
 Clears an accumulator.
virtual void updateAccumulator (TupleDatum &accumulatorDatum, TupleData const &inputTuple)
 Updates an accumulator with a new input tuple.
virtual void computeOutput (TupleDatum &outputDatum, TupleDatum const &accumulatorDatum)
 Computes an output based on accumulator state.
virtual void initAccumulator (TupleDatum &accumulatorDatumDest, TupleData const &inputTuple)
 Initializes a new accumulator datum from an input tuple.
virtual void initAccumulator (TupleDatum &accumulatorDatumSrc, TupleDatum &accumulatorDatumDest)
 Initializes a new accumulator datum from an existing accumulator datum.
virtual void updateAccumulator (TupleDatum &accumulatorDatumSrc, TupleDatum &accumulatorDatumDest, TupleData const &inputTuple)
 Computes a new accumulator from an existing accumulator dataum and a new input tuple.

Static Public Member Functions

static AggComputer * newAggComputer (AggFunction aggFunction, TupleAttributeDescriptor const *pAttrDesc)
 Virtual constructor.

Protected Attributes

int iInputAttr

Private Attributes

AggComputer * pComputer

Detailed Description

LbmRepeatingAggComputer is an aggregate computer that wraps another aggregate computer.

Its input is expected to be bitmap tuples. On an update, it counts bits from the segment data field (expected to be the last one) and applies the update multiple times, once for each bit set in the segment data field.

Definition at line 37 of file LbmSortedAggExecStream.cpp.

Constructor & Destructor Documentation

LbmRepeatingAggComputer::LbmRepeatingAggComputer ( AggComputer *  pComputer  )  [explicit]

Definition at line 73 of file LbmSortedAggExecStream.cpp.

References pComputer.

00075 {
00076     this->pComputer = pComputer;
00077 }

Member Function Documentation

void LbmRepeatingAggComputer::setInputAttrIndex ( uint  iInputAttrIndex  )  [virtual]

Sets the attribute index from which this computer should read input values in source tuples.

Parameters:
iInputAttrIndex 0-based tuple attribute index

Reimplemented from AggComputer.

Definition at line 79 of file LbmSortedAggExecStream.cpp.

References pComputer, and AggComputer::setInputAttrIndex().

00080 {
00081     AggComputer::setInputAttrIndex(iInputAttrIndex);
00082     pComputer->setInputAttrIndex(iInputAttrIndex);
00083 }

void LbmRepeatingAggComputer::clearAccumulator ( TupleDatum &  accumulatorDatum  )  [virtual]

Clears an accumulator.

Parameters:
accumulatorDatum in-memory value to be cleared

Implements AggComputer.

Definition at line 85 of file LbmSortedAggExecStream.cpp.

References AggComputer::clearAccumulator(), and pComputer.

00087 {
00088     pComputer->clearAccumulator(accumulatorDatum);
00089 }

void LbmRepeatingAggComputer::updateAccumulator ( TupleDatum &  accumulatorDatum,
TupleData const &  inputTuple 
) [virtual]

Updates an accumulator with a new input tuple.

Parameters:
accumulatorDatum in-memory value to be updated
inputTuple source for update; no references to this data should be retained after this method returns

Implements AggComputer.

Definition at line 91 of file LbmSortedAggExecStream.cpp.

References LbmByteSegment::countBits(), pComputer, and AggComputer::updateAccumulator().

00094 {
00095     // segment data should be contained in the last field
00096     TupleDatum segmentDatum = inputTuple[inputTuple.size() - 1];
00097     uint loops = LbmByteSegment::countBits(segmentDatum);
00098 
00099     for (uint i = 0; i < loops; i++) {
00100         pComputer->updateAccumulator(accumulatorDatum, inputTuple);
00101     }
00102 }

void LbmRepeatingAggComputer::computeOutput ( TupleDatum &  outputDatum,
TupleDatum const &  accumulatorDatum 
) [virtual]

Computes an output based on accumulator state.

Parameters:
outputDatum receives reference to computed output in preparation for marshalling result
accumulatorDatum final in-memory accumulator state

Implements AggComputer.

Definition at line 104 of file LbmSortedAggExecStream.cpp.

References AggComputer::computeOutput(), and pComputer.

00107 {
00108     pComputer->computeOutput(outputDatum, accumulatorDatum);
00109 }

void LbmRepeatingAggComputer::initAccumulator ( TupleDatum &  accumulatorDatumDest,
TupleData const &  inputTuple 
) [virtual]

Initializes a new accumulator datum from an input tuple.

Parameters:
accumulatorDatumDest in-memory value to be updated. Memory needs to be associated with this datum by the caller.
inputTuple source for update; no references to this data should be retained after this method returns

Implements AggComputer.

Definition at line 111 of file LbmSortedAggExecStream.cpp.

00114 {
00115     // sorted aggregates never use this call
00116     assert(false);
00117 }

void LbmRepeatingAggComputer::initAccumulator ( TupleDatum &  accumulatorDatumSrc,
TupleDatum &  accumulatorDatumDest 
) [virtual]

Initializes a new accumulator datum from an existing accumulator datum.

Parameters:
accumulatorDatumSrc the existing accumulator datum
accumulatorDatumDest the new accumulator datum. Memory needs to be associated with this datum by the caller.

Implements AggComputer.

Definition at line 119 of file LbmSortedAggExecStream.cpp.

00122 {
00123     // sorted aggregates never use this call
00124     assert(false);
00125 }

void LbmRepeatingAggComputer::updateAccumulator ( TupleDatum &  accumulatorDatumSrc,
TupleDatum &  accumulatorDatumDest,
TupleData const &  inputTuple 
) [virtual]

Computes a new accumulator from an existing accumulator dataum and a new input tuple.

Parameters:
accumulatorDatumSrc the existing accumulator datum
accumulatorDatumDest the new accumulator datum. memory needs to be associated with this datum by the caller.
inputTuple source for update; no references to this data should be retained after this method returns

Implements AggComputer.

Definition at line 127 of file LbmSortedAggExecStream.cpp.

00131 {
00132     // sorted aggregates never use this call
00133     assert(false);
00134 }

AggComputer * AggComputer::newAggComputer ( AggFunction  aggFunction,
TupleAttributeDescriptor const *  pAttrDesc 
) [static, inherited]

Virtual constructor.

Parameters:
aggFunction function for which to construct a computer
pAttrDesc descriptor for input attribute, or NULL for no input attribute (as in COUNT(*))

Definition at line 34 of file AggComputer.cpp.

References AGG_FUNC_COUNT, AGG_FUNC_MAX, AGG_FUNC_MIN, AGG_FUNC_SINGLE_VALUE, AGG_FUNC_SUM, STANDARD_TYPE_DOUBLE, STANDARD_TYPE_INT_16, STANDARD_TYPE_INT_32, STANDARD_TYPE_INT_64, STANDARD_TYPE_INT_8, STANDARD_TYPE_REAL, STANDARD_TYPE_UINT_16, STANDARD_TYPE_UINT_32, STANDARD_TYPE_UINT_64, and STANDARD_TYPE_UINT_8.

Referenced by SortedAggExecStream::newAggComputer(), and LhxAggExecStream::setAggComputers().

00037 {
00038     switch (aggFunction) {
00039     case AGG_FUNC_COUNT:
00040         if (pAttrDesc) {
00041             return new CountNullableAggComputer();
00042         } else {
00043             return new CountStarAggComputer();
00044         }
00045     case AGG_FUNC_SUM:
00046         assert(pAttrDesc);
00047         // TODO jvs 6-Oct-2005:  gotta use some of that template
00048         // metaprogramming mumbo jumbo to get rid of this
00049         switch (pAttrDesc->pTypeDescriptor->getOrdinal()) {
00050         case STANDARD_TYPE_INT_8:
00051             return new SumAggComputer<int8_t>();
00052         case STANDARD_TYPE_UINT_8:
00053             return new SumAggComputer<uint8_t>();
00054         case STANDARD_TYPE_INT_16:
00055             return new SumAggComputer<int16_t>();
00056         case STANDARD_TYPE_UINT_16:
00057             return new SumAggComputer<uint16_t>();
00058         case STANDARD_TYPE_INT_32:
00059             return new SumAggComputer<int32_t>();
00060         case STANDARD_TYPE_UINT_32:
00061             return new SumAggComputer<uint32_t>();
00062         case STANDARD_TYPE_INT_64:
00063             return new SumAggComputer<int64_t>();
00064         case STANDARD_TYPE_UINT_64:
00065             return new SumAggComputer<uint64_t>();
00066         case STANDARD_TYPE_REAL:
00067             return new SumAggComputer<float>();
00068         case STANDARD_TYPE_DOUBLE:
00069             return new SumAggComputer<double>();
00070         }
00071     case AGG_FUNC_MIN:
00072     case AGG_FUNC_MAX:
00073     case AGG_FUNC_SINGLE_VALUE:
00074         assert(pAttrDesc);
00075         return new ExtremeAggComputer(aggFunction, *pAttrDesc);
00076     }
00077     permAssert(false);
00078 }

Member Data Documentation

AggComputer* LbmRepeatingAggComputer::pComputer [private]

Definition at line 39 of file LbmSortedAggExecStream.cpp.

Referenced by clearAccumulator(), computeOutput(), LbmRepeatingAggComputer(), setInputAttrIndex(), and updateAccumulator().

int AggComputer::iInputAttr [protected, inherited]

Definition at line 50 of file AggComputer.h.

Referenced by AggComputer::AggComputer(), CountStarAggComputer::clearAccumulator(), SumAggComputer< T >::initAccumulator(), ExtremeAggComputer::initAccumulator(), CountNullableAggComputer::initAccumulator(), CountStarAggComputer::initAccumulator(), AggComputer::setInputAttrIndex(), SumAggComputer< T >::updateAccumulator(), ExtremeAggComputer::updateAccumulator(), and CountNullableAggComputer::updateAccumulator().

The documentation for this class was generated from the following file:
Generated on Mon Jun 22 04:00:35 2009 for Fennel by  doxygen 1.5.1