LhxHashTableReader Class Reference

#include <LhxHashTable.h>

List of all members.

Public Member Functions
void	init (LhxHashTable *hashTableInit, LhxHashInfo const &hashInfo, uint buildInputIndex)
	Initialize the hash table reader.
void	bindKey (PBuffer key)
	Bind this reader to a certain key.
void	bindUnMatched ()
	Bind this reader to unmatched keys.
bool	getNext (TupleData &outputTuple)
	Get the next outputTuple.
LhxHashTable *	getHashTable ()
Private Member Functions
bool	advanceSlot ()
	Locate the next slot to produce tuples from.
bool	advanceKey ()
	Locate the next key to produce tuples from.
bool	advanceData ()
	Locate the next data to produce tuples from.
void	produceTuple (TupleData &outputTuple)
	Produce the curKey + curData into outputTuple.
void	bind (PBuffer key)
	Helper function for bindKey() and bindUnMatched().
Private Attributes
LhxHashTable *	hashTable
	Underlying hash table to read from.
bool	isGroupBy
	Marks if this hash table is built for aggregation.
PBuffer *	curSlot
	Current read location.
PBuffer	curKey
PBuffer	curData
PBuffer	boundKey
	If not NULL, only read tuple with matching keys.
bool	returnUnMatched
	If true, only return tuples with unmatched keys.
bool	isPositioned
	Whether reader is positioned.
LhxHashKeyAccessor	hashKeyAccessor
	Accessors for the content of this hash table.
LhxHashDataAccessor	hashDataAccessor
TupleProjection	keyColsAndAggsProj
	Fields in the outputTuple that will hold keyCols and Aggs, and data columns.
TupleProjection	dataProj

---

Detailed Description

Definition at line 908 of file LhxHashTable.h.

---

Member Function Documentation

bool LhxHashTableReader::advanceSlot

(

)

[private]

Locate the next slot to produce tuples from.

Set up the curKey and curData pointers.

Returns:

false if there is no more slots with keys.

Definition at line 1019 of file LhxHashTable.cpp.

References boundKey, curData, curKey, curSlot, LhxHashKeyAccessor::getFirstData(), LhxHashTable::getFirstSlot(), LhxHashNodeAccessor::getNext(), LhxHashTable::getNextSlot(), hashDataAccessor, hashKeyAccessor, hashTable, isGroupBy, LhxHashKeyAccessor::isMatched(), isPositioned, returnUnMatched, LhxHashDataAccessor::setCurrent(), and LhxHashKeyAccessor::setCurrent().

Referenced by getNext().

{
    if (!boundKey) {
        curKey = NULL;

        if (!isPositioned) {
            curSlot = hashTable->getFirstSlot();
        } else {
            curSlot = hashTable->getNextSlot(curSlot);
        }

        if (curSlot && *curSlot) {
            curKey = *curSlot;
            if (returnUnMatched) {
                /*
                 * Look at all the keys in this slot.
                 */
                hashKeyAccessor.setCurrent(*curSlot, true);

                /*
                 * Only return unmatched keys
                 */
                while (hashKeyAccessor.isMatched()) {
                    curKey = hashKeyAccessor.getNext();
                    if (!curKey) {
                        curSlot = hashTable->getNextSlot(curSlot);
                        if (curSlot) {
                            curKey = *curSlot;
                        } else {
                            curKey = NULL;
                        }
                    }

                    if (curKey) {
                        hashKeyAccessor.setCurrent(curKey, true);
                    } else {
                        /*
                         * Reached the end of the slot list.
                         */
                        break;
                    }
                }
            }
        }

        if (!curKey) {
            /*
             * Cound not find any slot with fitting key.
             */
            return false;
        }
    }

    hashKeyAccessor.setCurrent(curKey, true);

    if (!isGroupBy) {
        curData = hashKeyAccessor.getFirstData();
        assert(curData);
        hashDataAccessor.setCurrent(curData, true);
    }

    return true;
}

bool LhxHashTableReader::advanceKey

(

)

[private]

Locate the next key to produce tuples from.

Set up the curKey and curData pointers.

Returns:

false if there is no more keys in the same slot.

Definition at line 1083 of file LhxHashTable.cpp.

References curData, curKey, LhxHashKeyAccessor::getFirstData(), LhxHashNodeAccessor::getNext(), hashDataAccessor, hashKeyAccessor, isGroupBy, LhxHashKeyAccessor::isMatched(), returnUnMatched, LhxHashDataAccessor::setCurrent(), and LhxHashKeyAccessor::setCurrent().

Referenced by getNext().

{
    while ((curKey = hashKeyAccessor.getNext())) {
        if (!returnUnMatched) {
            break;
        } else {
            hashKeyAccessor.setCurrent(curKey, true);
            if (!hashKeyAccessor.isMatched()) {
                break;
            }
        }
    }

    if (curKey) {
        hashKeyAccessor.setCurrent(curKey, true);
        if (!isGroupBy) {
            curData = hashKeyAccessor.getFirstData();
            assert(curData);
            hashDataAccessor.setCurrent(curData, true);
        }
        return true;
    } else {
        return false;
    }
}

bool LhxHashTableReader::advanceData

(

)

[private]

Locate the next data to produce tuples from.

Set up the curData pointers.

Returns:

false if there is no more data with the same key.

Definition at line 1109 of file LhxHashTable.cpp.

References curData, LhxHashNodeAccessor::getNext(), hashDataAccessor, isGroupBy, and LhxHashDataAccessor::setCurrent().

Referenced by getNext().

{
    if (isGroupBy) {
        return false;
    }

    curData = hashDataAccessor.getNext();
    if (curData) {
        hashDataAccessor.setCurrent(curData, true);
        return true;
    } else {
        return false;
    }
}

void LhxHashTableReader::produceTuple

(

TupleData &

outputTuple

)

[private]

Produce the curKey + curData into outputTuple.

Definition at line 1124 of file LhxHashTable.cpp.

References dataProj, hashDataAccessor, hashKeyAccessor, isGroupBy, keyColsAndAggsProj, LhxHashDataAccessor::unpack(), and LhxHashKeyAccessor::unpack().

Referenced by getNext().

{
    hashKeyAccessor.unpack(outputTuple, keyColsAndAggsProj);
    if (!isGroupBy) {
        hashDataAccessor.unpack(outputTuple, dataProj);
    }
}

void LhxHashTableReader::bind

(

PBuffer

key

)

[inline, private]

Helper function for bindKey() and bindUnMatched().

Definition at line 1313 of file LhxHashTable.h.

References boundKey, curData, curKey, curSlot, and isPositioned.

Referenced by bindKey(), and bindUnMatched().

{
    boundKey = curKey = key;
    curSlot = NULL;
    curData = NULL;
    isPositioned = false;
}

void LhxHashTableReader::init	(	LhxHashTable *	hashTableInit,
		LhxHashInfo const &	hashInfo,
		uint	buildInputIndex
	)

Initialize the hash table reader.

Parameters:

[in]	hashTableInit	the underlying hash table to read from
[in]	hashInfo
[in]	buildInputIndex	which input is the build side.

Definition at line 1132 of file LhxHashTable.cpp.

References LhxHashInfo::aggsProj, bindKey(), LhxHashInfo::dataProj, dataProj, hashDataAccessor, hashKeyAccessor, hashTable, LhxHashDataAccessor::init(), LhxHashKeyAccessor::init(), LhxHashInfo::inputDesc, isGroupBy, LhxHashTable::isHashGroupBy(), keyColsAndAggsProj, and LhxHashInfo::keyProj.

Referenced by LhxJoinExecStream::execute(), LhxAggExecStream::execute(), LhxPartitionWriter::open(), LhxJoinExecStream::open(), LhxAggExecStream::open(), LhxHashTableTest::testInsert(), and LhxHashTableTest::writeHashTable().

{
    /*
     * The last input is the build side.
     */
    TupleDescriptor const &outputTupleDesc =
        hashInfo.inputDesc[buildInputIndex];
    TupleProjection const &keyColsProj = hashInfo.keyProj[buildInputIndex];
    TupleProjection const &aggsProj = hashInfo.aggsProj;

    dataProj = hashInfo.dataProj[buildInputIndex];

    /*
     * These steps initialize the keyColsProjInKey and aggsProjInKey which are
     * based on the new keyColsAndAggs tuple.
     */
    TupleDescriptor keyDesc;
    TupleDescriptor dataDesc;
    TupleProjection keyColsProjInKey;
    TupleProjection aggsProjInKey;
    uint keyCount = keyColsProj.size();
    uint i;

    for (i = 0; i < keyCount; i++) {
        keyDesc.push_back(outputTupleDesc[keyColsProj[i]]);
        keyColsProjInKey.push_back(i);
    }

    keyColsAndAggsProj = keyColsProj;
    uint aggsProjSize = aggsProj.size();

    for (i = 0; i < aggsProjSize; i ++) {
        keyColsAndAggsProj.push_back(aggsProj[i]);
        keyDesc.push_back(outputTupleDesc[aggsProj[i]]);
        aggsProjInKey.push_back(i + keyCount);
    }

    hashKeyAccessor.init(keyDesc, keyColsProjInKey, aggsProjInKey);

    for (i = 0; i < dataProj.size(); i++) {
        dataDesc.push_back(outputTupleDesc[dataProj[i]]);
    }

    hashDataAccessor.init(dataDesc);

    hashTable = hashTableInit;
    isGroupBy = hashTable->isHashGroupBy();

    /*
     * By default, this reader is not associated with any key, i.e. it covers
     * the whole hash table.
     */
    bindKey(NULL);
}

void LhxHashTableReader::bindKey

(

PBuffer

key

)

[inline]

Bind this reader to a certain key.

Only tuples with the same key are returned.

Parameters:

[in]

key

key node to bind this reader to.

Definition at line 1321 of file LhxHashTable.h.

References bind(), and returnUnMatched.

Referenced by LhxPartitionWriter::aggAndMarshalTuple(), LhxJoinExecStream::execute(), init(), LhxPartitionInfo::open(), LhxHashTableTest::testInsert(), and LhxHashTableTest::writeHashTable().

{
    returnUnMatched = false;
    bind(key);
}

void LhxHashTableReader::bindUnMatched

(

)

[inline]

Bind this reader to unmatched keys.

Only rows with unmatched keys will be returned.

Definition at line 1327 of file LhxHashTable.h.

References bind(), and returnUnMatched.

Referenced by LhxJoinExecStream::execute().

{
    returnUnMatched = true;
    bind(NULL);
}

bool LhxHashTableReader::getNext

(

TupleData &

outputTuple

)

Get the next outputTuple.

Parameters:

[out]

outputTuple

Returns:

false if no more tuples to output.

Definition at line 1190 of file LhxHashTable.cpp.

References advanceData(), advanceKey(), advanceSlot(), boundKey, isPositioned, produceTuple(), and returnUnMatched.

Referenced by LhxPartitionWriter::aggAndMarshalTuple(), LhxPartitionWriter::close(), LhxJoinExecStream::execute(), LhxAggExecStream::execute(), LhxHashTableTest::testInsert(), and LhxHashTableTest::writeHashTable().

{
    if (!isPositioned) {
        assert (!(boundKey && returnUnMatched));

        /*
         * Position at the first qualifying key of the first slot.
         */
        if (!advanceSlot()) {
            /*
             * Nothing to output.
             */
            return false;
        }
        produceTuple(outputTuple);
        isPositioned = true;
        return true;
    }

    if (advanceData()) {
        produceTuple(outputTuple);
        return true;
    } else {
        if (boundKey) {
            /*
             * End of data for this key.
             */
            return false;
        } else {
            /*
             * Get the next key in the same slot.
             */
            if (advanceKey()) {
                produceTuple(outputTuple);
                return true;
            } else {
                /*
                 * End of list for keys in the same slot..
                 * Start the next slot.
                 */
                if (advanceSlot()) {
                    produceTuple(outputTuple);
                    return true;
                } else {
                    return false;
                }
            }
        }
    }
}

LhxHashTable * LhxHashTableReader::getHashTable

(

)

[inline]

Definition at line 1308 of file LhxHashTable.h.

References hashTable.

{
    return hashTable;
}

---

Member Data Documentation

LhxHashTable* LhxHashTableReader::hashTable [private]

Underlying hash table to read from.

Definition at line 913 of file LhxHashTable.h.

Referenced by advanceSlot(), getHashTable(), and init().

bool LhxHashTableReader::isGroupBy [private]

Marks if this hash table is built for aggregation.

Aggregating hash table only contains keys (group by keys plus aggregates) and does not contain data portion. The reader behavior will thus be different from reading a hash table built for joins.

Definition at line 922 of file LhxHashTable.h.

Referenced by advanceData(), advanceKey(), advanceSlot(), init(), and produceTuple().

PBuffer* LhxHashTableReader::curSlot [private]

Current read location.

Definition at line 927 of file LhxHashTable.h.

Referenced by advanceSlot(), and bind().

PBuffer LhxHashTableReader::curKey [private]

Definition at line 928 of file LhxHashTable.h.

Referenced by advanceKey(), advanceSlot(), and bind().

PBuffer LhxHashTableReader::curData [private]

Definition at line 929 of file LhxHashTable.h.

Referenced by advanceData(), advanceKey(), advanceSlot(), and bind().

PBuffer LhxHashTableReader::boundKey [private]

If not NULL, only read tuple with matching keys.

Not compatible with returnUnMatched

Definition at line 937 of file LhxHashTable.h.

Referenced by advanceSlot(), bind(), and getNext().

bool LhxHashTableReader::returnUnMatched [private]

If true, only return tuples with unmatched keys.

This is not compatible with boundKey.

Definition at line 943 of file LhxHashTable.h.

Referenced by advanceKey(), advanceSlot(), bindKey(), bindUnMatched(), and getNext().

bool LhxHashTableReader::isPositioned [private]

Whether reader is positioned.

Definition at line 948 of file LhxHashTable.h.

Referenced by advanceSlot(), bind(), and getNext().

LhxHashKeyAccessor LhxHashTableReader::hashKeyAccessor [private]

Accessors for the content of this hash table.

Definition at line 953 of file LhxHashTable.h.

Referenced by advanceKey(), advanceSlot(), init(), and produceTuple().

LhxHashDataAccessor LhxHashTableReader::hashDataAccessor [private]

Definition at line 954 of file LhxHashTable.h.

Referenced by advanceData(), advanceKey(), advanceSlot(), init(), and produceTuple().

TupleProjection LhxHashTableReader::keyColsAndAggsProj [private]

Fields in the outputTuple that will hold keyCols and Aggs, and data columns.

output tuple should have the same shape as outputTupleDesc used in the init() method.

Definition at line 963 of file LhxHashTable.h.

Referenced by init(), and produceTuple().

TupleProjection LhxHashTableReader::dataProj [private]

Definition at line 964 of file LhxHashTable.h.

Referenced by init(), and produceTuple().

---

The documentation for this class was generated from the following files:

• /home/pub/open/dev/fennel/hashexe/LhxHashTable.h
• /home/pub/open/dev/fennel/hashexe/LhxHashTable.cpp

---