Examples with BooleanColumnStatsData org.apache.hadoop.hive.metastore.api.BooleanColumnStatsData used on opensource projects

Example 11 with BooleanColumnStatsData

use of org.apache.hadoop.hive.metastore.api.BooleanColumnStatsData in project hive by apache.

the class TestHBaseStore method booleanTableStatistics.

// Due to the way our mock stuff works, we can only insert one column at a time, so we'll test
// each stat type separately.  We'll test them together in the integration tests.
@Test
public void booleanTableStatistics() throws Exception {
    // Add a boolean table stats for BOOLEAN_COL to DB
    // Because of the way our mock implementation works we actually need to not create the table
    // before we set statistics on it.
    ColumnStatistics stats = new ColumnStatistics();
    // Get a default ColumnStatisticsDesc for table level stats
    ColumnStatisticsDesc desc = getMockTblColStatsDesc();
    stats.setStatsDesc(desc);
    // Get one of the pre-created ColumnStatisticsObj
    ColumnStatisticsObj obj = booleanColStatsObjs.get(0);
    BooleanColumnStatsData boolData = obj.getStatsData().getBooleanStats();
    // Add to DB
    stats.addToStatsObj(obj);
    store.updateTableColumnStatistics(stats);
    // Get from DB
    ColumnStatistics statsFromDB = store.getTableColumnStatistics(DB, TBL, Arrays.asList(BOOLEAN_COL));
    // Compare ColumnStatisticsDesc
    Assert.assertEquals(desc.getLastAnalyzed(), statsFromDB.getStatsDesc().getLastAnalyzed());
    Assert.assertEquals(DB, statsFromDB.getStatsDesc().getDbName());
    Assert.assertEquals(TBL, statsFromDB.getStatsDesc().getTableName());
    Assert.assertTrue(statsFromDB.getStatsDesc().isIsTblLevel());
    // Compare ColumnStatisticsObj
    Assert.assertEquals(1, statsFromDB.getStatsObjSize());
    ColumnStatisticsObj objFromDB = statsFromDB.getStatsObj().get(0);
    ColumnStatisticsData dataFromDB = objFromDB.getStatsData();
    // Compare ColumnStatisticsData
    Assert.assertEquals(ColumnStatisticsData._Fields.BOOLEAN_STATS, dataFromDB.getSetField());
    // Compare BooleanColumnStatsData
    BooleanColumnStatsData boolDataFromDB = dataFromDB.getBooleanStats();
    Assert.assertEquals(boolData.getNumTrues(), boolDataFromDB.getNumTrues());
    Assert.assertEquals(boolData.getNumFalses(), boolDataFromDB.getNumFalses());
    Assert.assertEquals(boolData.getNumNulls(), boolDataFromDB.getNumNulls());
}

Example 12 with BooleanColumnStatsData

use of org.apache.hadoop.hive.metastore.api.BooleanColumnStatsData in project hive by apache.

the class TestHBaseStore method mockBooleanStats.

private static ColumnStatisticsObj mockBooleanStats(int i) {
    long trues = 37 + 100 * i;
    long falses = 12 + 50 * i;
    long nulls = 2 + i;
    ColumnStatisticsObj colStatsObj = new ColumnStatisticsObj();
    colStatsObj.setColName(BOOLEAN_COL);
    colStatsObj.setColType(BOOLEAN_TYPE);
    ColumnStatisticsData data = new ColumnStatisticsData();
    BooleanColumnStatsData boolData = new BooleanColumnStatsData();
    boolData.setNumTrues(trues);
    boolData.setNumFalses(falses);
    boolData.setNumNulls(nulls);
    data.setBooleanStats(boolData);
    colStatsObj.setStatsData(data);
    return colStatsObj;
}

Example 13 with BooleanColumnStatsData

use of org.apache.hadoop.hive.metastore.api.BooleanColumnStatsData in project hive by apache.

the class StatObjectConverter method fillColumnStatisticsData.

// DB
public static void fillColumnStatisticsData(String colType, ColumnStatisticsData data, Object llow, Object lhigh, Object dlow, Object dhigh, Object declow, Object dechigh, Object nulls, Object dist, Object avglen, Object maxlen, Object trues, Object falses, Object avgLong, Object avgDouble, Object avgDecimal, Object sumDist, boolean useDensityFunctionForNDVEstimation, double ndvTuner) throws MetaException {
    colType = colType.toLowerCase();
    if (colType.equals("boolean")) {
        BooleanColumnStatsData boolStats = new BooleanColumnStatsData();
        boolStats.setNumFalses(MetaStoreDirectSql.extractSqlLong(falses));
        boolStats.setNumTrues(MetaStoreDirectSql.extractSqlLong(trues));
        boolStats.setNumNulls(MetaStoreDirectSql.extractSqlLong(nulls));
        data.setBooleanStats(boolStats);
    } else if (colType.equals("string") || colType.startsWith("varchar") || colType.startsWith("char")) {
        StringColumnStatsDataInspector stringStats = new StringColumnStatsDataInspector();
        stringStats.setNumNulls(MetaStoreDirectSql.extractSqlLong(nulls));
        stringStats.setAvgColLen(MetaStoreDirectSql.extractSqlDouble(avglen));
        stringStats.setMaxColLen(MetaStoreDirectSql.extractSqlLong(maxlen));
        stringStats.setNumDVs(MetaStoreDirectSql.extractSqlLong(dist));
        data.setStringStats(stringStats);
    } else if (colType.equals("binary")) {
        BinaryColumnStatsData binaryStats = new BinaryColumnStatsData();
        binaryStats.setNumNulls(MetaStoreDirectSql.extractSqlLong(nulls));
        binaryStats.setAvgColLen(MetaStoreDirectSql.extractSqlDouble(avglen));
        binaryStats.setMaxColLen(MetaStoreDirectSql.extractSqlLong(maxlen));
        data.setBinaryStats(binaryStats);
    } else if (colType.equals("bigint") || colType.equals("int") || colType.equals("smallint") || colType.equals("tinyint") || colType.equals("timestamp")) {
        LongColumnStatsDataInspector longStats = new LongColumnStatsDataInspector();
        longStats.setNumNulls(MetaStoreDirectSql.extractSqlLong(nulls));
        if (lhigh != null) {
            longStats.setHighValue(MetaStoreDirectSql.extractSqlLong(lhigh));
        }
        if (llow != null) {
            longStats.setLowValue(MetaStoreDirectSql.extractSqlLong(llow));
        }
        long lowerBound = MetaStoreDirectSql.extractSqlLong(dist);
        long higherBound = MetaStoreDirectSql.extractSqlLong(sumDist);
        long rangeBound = Long.MAX_VALUE;
        if (lhigh != null && llow != null) {
            rangeBound = MetaStoreDirectSql.extractSqlLong(lhigh) - MetaStoreDirectSql.extractSqlLong(llow) + 1;
        }
        long estimation;
        if (useDensityFunctionForNDVEstimation && lhigh != null && llow != null && avgLong != null && MetaStoreDirectSql.extractSqlDouble(avgLong) != 0.0) {
            // We have estimation, lowerbound and higherbound. We use estimation if
            // it is between lowerbound and higherbound.
            estimation = MetaStoreDirectSql.extractSqlLong((MetaStoreDirectSql.extractSqlLong(lhigh) - MetaStoreDirectSql.extractSqlLong(llow)) / MetaStoreDirectSql.extractSqlDouble(avgLong));
            if (estimation < lowerBound) {
                estimation = lowerBound;
            } else if (estimation > higherBound) {
                estimation = higherBound;
            }
        } else {
            estimation = (long) (lowerBound + (higherBound - lowerBound) * ndvTuner);
        }
        estimation = Math.min(estimation, rangeBound);
        longStats.setNumDVs(estimation);
        data.setLongStats(longStats);
    } else if (colType.equals("date")) {
        DateColumnStatsDataInspector dateStats = new DateColumnStatsDataInspector();
        dateStats.setNumNulls(MetaStoreDirectSql.extractSqlLong(nulls));
        if (lhigh != null) {
            dateStats.setHighValue(new Date(MetaStoreDirectSql.extractSqlLong(lhigh)));
        }
        if (llow != null) {
            dateStats.setLowValue(new Date(MetaStoreDirectSql.extractSqlLong(llow)));
        }
        long lowerBound = MetaStoreDirectSql.extractSqlLong(dist);
        long higherBound = MetaStoreDirectSql.extractSqlLong(sumDist);
        long rangeBound = Long.MAX_VALUE;
        if (lhigh != null && llow != null) {
            rangeBound = MetaStoreDirectSql.extractSqlLong(lhigh) - MetaStoreDirectSql.extractSqlLong(llow) + 1;
        }
        long estimation;
        if (useDensityFunctionForNDVEstimation && lhigh != null && llow != null && avgLong != null && MetaStoreDirectSql.extractSqlDouble(avgLong) != 0.0) {
            // We have estimation, lowerbound and higherbound. We use estimation if
            // it is between lowerbound and higherbound.
            estimation = MetaStoreDirectSql.extractSqlLong((MetaStoreDirectSql.extractSqlLong(lhigh) - MetaStoreDirectSql.extractSqlLong(llow)) / MetaStoreDirectSql.extractSqlDouble(avgLong));
            if (estimation < lowerBound) {
                estimation = lowerBound;
            } else if (estimation > higherBound) {
                estimation = higherBound;
            }
        } else {
            estimation = (long) (lowerBound + (higherBound - lowerBound) * ndvTuner);
        }
        estimation = Math.min(estimation, rangeBound);
        dateStats.setNumDVs(estimation);
        data.setDateStats(dateStats);
    } else if (colType.equals("double") || colType.equals("float")) {
        DoubleColumnStatsDataInspector doubleStats = new DoubleColumnStatsDataInspector();
        doubleStats.setNumNulls(MetaStoreDirectSql.extractSqlLong(nulls));
        if (dhigh != null) {
            doubleStats.setHighValue(MetaStoreDirectSql.extractSqlDouble(dhigh));
        }
        if (dlow != null) {
            doubleStats.setLowValue(MetaStoreDirectSql.extractSqlDouble(dlow));
        }
        long lowerBound = MetaStoreDirectSql.extractSqlLong(dist);
        long higherBound = MetaStoreDirectSql.extractSqlLong(sumDist);
        if (useDensityFunctionForNDVEstimation && dhigh != null && dlow != null && avgDouble != null && MetaStoreDirectSql.extractSqlDouble(avgDouble) != 0.0) {
            long estimation = MetaStoreDirectSql.extractSqlLong((MetaStoreDirectSql.extractSqlLong(dhigh) - MetaStoreDirectSql.extractSqlLong(dlow)) / MetaStoreDirectSql.extractSqlDouble(avgDouble));
            if (estimation < lowerBound) {
                doubleStats.setNumDVs(lowerBound);
            } else if (estimation > higherBound) {
                doubleStats.setNumDVs(higherBound);
            } else {
                doubleStats.setNumDVs(estimation);
            }
        } else {
            doubleStats.setNumDVs((long) (lowerBound + (higherBound - lowerBound) * ndvTuner));
        }
        data.setDoubleStats(doubleStats);
    } else if (colType.startsWith("decimal")) {
        DecimalColumnStatsDataInspector decimalStats = new DecimalColumnStatsDataInspector();
        decimalStats.setNumNulls(MetaStoreDirectSql.extractSqlLong(nulls));
        Decimal low = null;
        Decimal high = null;
        BigDecimal blow = null;
        BigDecimal bhigh = null;
        if (dechigh instanceof BigDecimal) {
            bhigh = (BigDecimal) dechigh;
            high = new Decimal(ByteBuffer.wrap(bhigh.unscaledValue().toByteArray()), (short) bhigh.scale());
        } else if (dechigh instanceof String) {
            bhigh = new BigDecimal((String) dechigh);
            high = createThriftDecimal((String) dechigh);
        }
        decimalStats.setHighValue(high);
        if (declow instanceof BigDecimal) {
            blow = (BigDecimal) declow;
            low = new Decimal(ByteBuffer.wrap(blow.unscaledValue().toByteArray()), (short) blow.scale());
        } else if (dechigh instanceof String) {
            blow = new BigDecimal((String) declow);
            low = createThriftDecimal((String) declow);
        }
        decimalStats.setLowValue(low);
        long lowerBound = MetaStoreDirectSql.extractSqlLong(dist);
        long higherBound = MetaStoreDirectSql.extractSqlLong(sumDist);
        if (useDensityFunctionForNDVEstimation && dechigh != null && declow != null && avgDecimal != null && MetaStoreDirectSql.extractSqlDouble(avgDecimal) != 0.0) {
            long estimation = MetaStoreDirectSql.extractSqlLong(MetaStoreDirectSql.extractSqlLong(bhigh.subtract(blow).floatValue() / MetaStoreDirectSql.extractSqlDouble(avgDecimal)));
            if (estimation < lowerBound) {
                decimalStats.setNumDVs(lowerBound);
            } else if (estimation > higherBound) {
                decimalStats.setNumDVs(higherBound);
            } else {
                decimalStats.setNumDVs(estimation);
            }
        } else {
            decimalStats.setNumDVs((long) (lowerBound + (higherBound - lowerBound) * ndvTuner));
        }
        data.setDecimalStats(decimalStats);
    }
}

Example 14 with BooleanColumnStatsData

use of org.apache.hadoop.hive.metastore.api.BooleanColumnStatsData in project hive by apache.

the class StatObjectConverter method convertToMPartitionColumnStatistics.

public static MPartitionColumnStatistics convertToMPartitionColumnStatistics(MPartition partition, ColumnStatisticsDesc statsDesc, ColumnStatisticsObj statsObj) throws MetaException, NoSuchObjectException {
    if (statsDesc == null || statsObj == null) {
        return null;
    }
    MPartitionColumnStatistics mColStats = new MPartitionColumnStatistics();
    mColStats.setPartition(partition);
    mColStats.setDbName(statsDesc.getDbName());
    mColStats.setTableName(statsDesc.getTableName());
    mColStats.setPartitionName(statsDesc.getPartName());
    mColStats.setLastAnalyzed(statsDesc.getLastAnalyzed());
    mColStats.setColName(statsObj.getColName());
    mColStats.setColType(statsObj.getColType());
    if (statsObj.getStatsData().isSetBooleanStats()) {
        BooleanColumnStatsData boolStats = statsObj.getStatsData().getBooleanStats();
        mColStats.setBooleanStats(boolStats.isSetNumTrues() ? boolStats.getNumTrues() : null, boolStats.isSetNumFalses() ? boolStats.getNumFalses() : null, boolStats.isSetNumNulls() ? boolStats.getNumNulls() : null);
    } else if (statsObj.getStatsData().isSetLongStats()) {
        LongColumnStatsData longStats = statsObj.getStatsData().getLongStats();
        mColStats.setLongStats(longStats.isSetNumNulls() ? longStats.getNumNulls() : null, longStats.isSetNumDVs() ? longStats.getNumDVs() : null, longStats.isSetBitVectors() ? longStats.getBitVectors() : null, longStats.isSetLowValue() ? longStats.getLowValue() : null, longStats.isSetHighValue() ? longStats.getHighValue() : null);
    } else if (statsObj.getStatsData().isSetDoubleStats()) {
        DoubleColumnStatsData doubleStats = statsObj.getStatsData().getDoubleStats();
        mColStats.setDoubleStats(doubleStats.isSetNumNulls() ? doubleStats.getNumNulls() : null, doubleStats.isSetNumDVs() ? doubleStats.getNumDVs() : null, doubleStats.isSetBitVectors() ? doubleStats.getBitVectors() : null, doubleStats.isSetLowValue() ? doubleStats.getLowValue() : null, doubleStats.isSetHighValue() ? doubleStats.getHighValue() : null);
    } else if (statsObj.getStatsData().isSetDecimalStats()) {
        DecimalColumnStatsData decimalStats = statsObj.getStatsData().getDecimalStats();
        String low = decimalStats.isSetLowValue() ? createJdoDecimalString(decimalStats.getLowValue()) : null;
        String high = decimalStats.isSetHighValue() ? createJdoDecimalString(decimalStats.getHighValue()) : null;
        mColStats.setDecimalStats(decimalStats.isSetNumNulls() ? decimalStats.getNumNulls() : null, decimalStats.isSetNumDVs() ? decimalStats.getNumDVs() : null, decimalStats.isSetBitVectors() ? decimalStats.getBitVectors() : null, low, high);
    } else if (statsObj.getStatsData().isSetStringStats()) {
        StringColumnStatsData stringStats = statsObj.getStatsData().getStringStats();
        mColStats.setStringStats(stringStats.isSetNumNulls() ? stringStats.getNumNulls() : null, stringStats.isSetNumDVs() ? stringStats.getNumDVs() : null, stringStats.isSetBitVectors() ? stringStats.getBitVectors() : null, stringStats.isSetMaxColLen() ? stringStats.getMaxColLen() : null, stringStats.isSetAvgColLen() ? stringStats.getAvgColLen() : null);
    } else if (statsObj.getStatsData().isSetBinaryStats()) {
        BinaryColumnStatsData binaryStats = statsObj.getStatsData().getBinaryStats();
        mColStats.setBinaryStats(binaryStats.isSetNumNulls() ? binaryStats.getNumNulls() : null, binaryStats.isSetMaxColLen() ? binaryStats.getMaxColLen() : null, binaryStats.isSetAvgColLen() ? binaryStats.getAvgColLen() : null);
    } else if (statsObj.getStatsData().isSetDateStats()) {
        DateColumnStatsData dateStats = statsObj.getStatsData().getDateStats();
        mColStats.setDateStats(dateStats.isSetNumNulls() ? dateStats.getNumNulls() : null, dateStats.isSetNumDVs() ? dateStats.getNumDVs() : null, dateStats.isSetBitVectors() ? dateStats.getBitVectors() : null, dateStats.isSetLowValue() ? dateStats.getLowValue().getDaysSinceEpoch() : null, dateStats.isSetHighValue() ? dateStats.getHighValue().getDaysSinceEpoch() : null);
    }
    return mColStats;
}

Example 15 with BooleanColumnStatsData

use of org.apache.hadoop.hive.metastore.api.BooleanColumnStatsData in project hive by apache.

the class StatObjectConverter method getPartitionColumnStatisticsObj.

public static ColumnStatisticsObj getPartitionColumnStatisticsObj(MPartitionColumnStatistics mStatsObj, boolean enableBitVector) {
    ColumnStatisticsObj statsObj = new ColumnStatisticsObj();
    statsObj.setColType(mStatsObj.getColType());
    statsObj.setColName(mStatsObj.getColName());
    String colType = mStatsObj.getColType().toLowerCase();
    ColumnStatisticsData colStatsData = new ColumnStatisticsData();
    if (colType.equals("boolean")) {
        BooleanColumnStatsData boolStats = new BooleanColumnStatsData();
        boolStats.setNumFalses(mStatsObj.getNumFalses());
        boolStats.setNumTrues(mStatsObj.getNumTrues());
        boolStats.setNumNulls(mStatsObj.getNumNulls());
        colStatsData.setBooleanStats(boolStats);
    } else if (colType.equals("string") || colType.startsWith("varchar") || colType.startsWith("char")) {
        StringColumnStatsDataInspector stringStats = new StringColumnStatsDataInspector();
        stringStats.setNumNulls(mStatsObj.getNumNulls());
        stringStats.setAvgColLen(mStatsObj.getAvgColLen());
        stringStats.setMaxColLen(mStatsObj.getMaxColLen());
        stringStats.setNumDVs(mStatsObj.getNumDVs());
        stringStats.setBitVectors((mStatsObj.getBitVector() == null || !enableBitVector) ? null : mStatsObj.getBitVector());
        colStatsData.setStringStats(stringStats);
    } else if (colType.equals("binary")) {
        BinaryColumnStatsData binaryStats = new BinaryColumnStatsData();
        binaryStats.setNumNulls(mStatsObj.getNumNulls());
        binaryStats.setAvgColLen(mStatsObj.getAvgColLen());
        binaryStats.setMaxColLen(mStatsObj.getMaxColLen());
        colStatsData.setBinaryStats(binaryStats);
    } else if (colType.equals("tinyint") || colType.equals("smallint") || colType.equals("int") || colType.equals("bigint") || colType.equals("timestamp")) {
        LongColumnStatsDataInspector longStats = new LongColumnStatsDataInspector();
        longStats.setNumNulls(mStatsObj.getNumNulls());
        if (mStatsObj.getLongHighValue() != null) {
            longStats.setHighValue(mStatsObj.getLongHighValue());
        }
        if (mStatsObj.getLongLowValue() != null) {
            longStats.setLowValue(mStatsObj.getLongLowValue());
        }
        longStats.setNumDVs(mStatsObj.getNumDVs());
        longStats.setBitVectors((mStatsObj.getBitVector() == null || !enableBitVector) ? null : mStatsObj.getBitVector());
        colStatsData.setLongStats(longStats);
    } else if (colType.equals("double") || colType.equals("float")) {
        DoubleColumnStatsDataInspector doubleStats = new DoubleColumnStatsDataInspector();
        doubleStats.setNumNulls(mStatsObj.getNumNulls());
        if (mStatsObj.getDoubleHighValue() != null) {
            doubleStats.setHighValue(mStatsObj.getDoubleHighValue());
        }
        if (mStatsObj.getDoubleLowValue() != null) {
            doubleStats.setLowValue(mStatsObj.getDoubleLowValue());
        }
        doubleStats.setNumDVs(mStatsObj.getNumDVs());
        doubleStats.setBitVectors((mStatsObj.getBitVector() == null || !enableBitVector) ? null : mStatsObj.getBitVector());
        colStatsData.setDoubleStats(doubleStats);
    } else if (colType.startsWith("decimal")) {
        DecimalColumnStatsDataInspector decimalStats = new DecimalColumnStatsDataInspector();
        decimalStats.setNumNulls(mStatsObj.getNumNulls());
        if (mStatsObj.getDecimalHighValue() != null) {
            decimalStats.setHighValue(createThriftDecimal(mStatsObj.getDecimalHighValue()));
        }
        if (mStatsObj.getDecimalLowValue() != null) {
            decimalStats.setLowValue(createThriftDecimal(mStatsObj.getDecimalLowValue()));
        }
        decimalStats.setNumDVs(mStatsObj.getNumDVs());
        decimalStats.setBitVectors((mStatsObj.getBitVector() == null || !enableBitVector) ? null : mStatsObj.getBitVector());
        colStatsData.setDecimalStats(decimalStats);
    } else if (colType.equals("date")) {
        DateColumnStatsDataInspector dateStats = new DateColumnStatsDataInspector();
        dateStats.setNumNulls(mStatsObj.getNumNulls());
        dateStats.setHighValue(new Date(mStatsObj.getLongHighValue()));
        dateStats.setLowValue(new Date(mStatsObj.getLongLowValue()));
        dateStats.setNumDVs(mStatsObj.getNumDVs());
        dateStats.setBitVectors((mStatsObj.getBitVector() == null || !enableBitVector) ? null : mStatsObj.getBitVector());
        colStatsData.setDateStats(dateStats);
    }
    statsObj.setStatsData(colStatsData);
    return statsObj;
}