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

Example 16 with BinaryColumnStatsData

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

the class HBaseUtils method statsForOneColumnFromProtoBuf.

private static ColumnStatisticsObj statsForOneColumnFromProtoBuf(ColumnStatistics partitionColumnStats, HbaseMetastoreProto.ColumnStats proto) throws IOException {
    ColumnStatisticsObj colStats = new ColumnStatisticsObj();
    long lastAnalyzed = proto.getLastAnalyzed();
    if (partitionColumnStats != null) {
        partitionColumnStats.getStatsDesc().setLastAnalyzed(Math.max(lastAnalyzed, partitionColumnStats.getStatsDesc().getLastAnalyzed()));
    }
    colStats.setColType(proto.getColumnType());
    colStats.setColName(proto.getColumnName());
    ColumnStatisticsData colData = new ColumnStatisticsData();
    if (proto.hasBoolStats()) {
        BooleanColumnStatsData boolData = new BooleanColumnStatsData();
        boolData.setNumTrues(proto.getBoolStats().getNumTrues());
        boolData.setNumFalses(proto.getBoolStats().getNumFalses());
        boolData.setNumNulls(proto.getNumNulls());
        colData.setBooleanStats(boolData);
    } else if (proto.hasLongStats()) {
        LongColumnStatsData longData = new LongColumnStatsData();
        if (proto.getLongStats().hasLowValue()) {
            longData.setLowValue(proto.getLongStats().getLowValue());
        }
        if (proto.getLongStats().hasHighValue()) {
            longData.setHighValue(proto.getLongStats().getHighValue());
        }
        longData.setNumNulls(proto.getNumNulls());
        longData.setNumDVs(proto.getNumDistinctValues());
        longData.setBitVectors(proto.getBitVectors());
        colData.setLongStats(longData);
    } else if (proto.hasDoubleStats()) {
        DoubleColumnStatsData doubleData = new DoubleColumnStatsData();
        if (proto.getDoubleStats().hasLowValue()) {
            doubleData.setLowValue(proto.getDoubleStats().getLowValue());
        }
        if (proto.getDoubleStats().hasHighValue()) {
            doubleData.setHighValue(proto.getDoubleStats().getHighValue());
        }
        doubleData.setNumNulls(proto.getNumNulls());
        doubleData.setNumDVs(proto.getNumDistinctValues());
        doubleData.setBitVectors(proto.getBitVectors());
        colData.setDoubleStats(doubleData);
    } else if (proto.hasStringStats()) {
        StringColumnStatsData stringData = new StringColumnStatsData();
        stringData.setMaxColLen(proto.getStringStats().getMaxColLength());
        stringData.setAvgColLen(proto.getStringStats().getAvgColLength());
        stringData.setNumNulls(proto.getNumNulls());
        stringData.setNumDVs(proto.getNumDistinctValues());
        stringData.setBitVectors(proto.getBitVectors());
        colData.setStringStats(stringData);
    } else if (proto.hasBinaryStats()) {
        BinaryColumnStatsData binaryData = new BinaryColumnStatsData();
        binaryData.setMaxColLen(proto.getBinaryStats().getMaxColLength());
        binaryData.setAvgColLen(proto.getBinaryStats().getAvgColLength());
        binaryData.setNumNulls(proto.getNumNulls());
        colData.setBinaryStats(binaryData);
    } else if (proto.hasDecimalStats()) {
        DecimalColumnStatsData decimalData = new DecimalColumnStatsData();
        if (proto.getDecimalStats().hasHighValue()) {
            Decimal hiVal = new Decimal();
            hiVal.setUnscaled(proto.getDecimalStats().getHighValue().getUnscaled().toByteArray());
            hiVal.setScale((short) proto.getDecimalStats().getHighValue().getScale());
            decimalData.setHighValue(hiVal);
        }
        if (proto.getDecimalStats().hasLowValue()) {
            Decimal loVal = new Decimal();
            loVal.setUnscaled(proto.getDecimalStats().getLowValue().getUnscaled().toByteArray());
            loVal.setScale((short) proto.getDecimalStats().getLowValue().getScale());
            decimalData.setLowValue(loVal);
        }
        decimalData.setNumNulls(proto.getNumNulls());
        decimalData.setNumDVs(proto.getNumDistinctValues());
        decimalData.setBitVectors(proto.getBitVectors());
        colData.setDecimalStats(decimalData);
    } else {
        throw new RuntimeException("Woh, bad.  Unknown stats type!");
    }
    colStats.setStatsData(colData);
    return colStats;
}

Example 17 with BinaryColumnStatsData

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

the class HBaseUtils method protoBufStatsForOneColumn.

private static HbaseMetastoreProto.ColumnStats protoBufStatsForOneColumn(ColumnStatistics partitionColumnStats, ColumnStatisticsObj colStats) throws IOException {
    HbaseMetastoreProto.ColumnStats.Builder builder = HbaseMetastoreProto.ColumnStats.newBuilder();
    if (partitionColumnStats != null) {
        builder.setLastAnalyzed(partitionColumnStats.getStatsDesc().getLastAnalyzed());
    }
    assert colStats.getColType() != null;
    builder.setColumnType(colStats.getColType());
    assert colStats.getColName() != null;
    builder.setColumnName(colStats.getColName());
    ColumnStatisticsData colData = colStats.getStatsData();
    switch(colData.getSetField()) {
        case BOOLEAN_STATS:
            BooleanColumnStatsData boolData = colData.getBooleanStats();
            builder.setNumNulls(boolData.getNumNulls());
            builder.setBoolStats(HbaseMetastoreProto.ColumnStats.BooleanStats.newBuilder().setNumTrues(boolData.getNumTrues()).setNumFalses(boolData.getNumFalses()).build());
            break;
        case LONG_STATS:
            LongColumnStatsData longData = colData.getLongStats();
            builder.setNumNulls(longData.getNumNulls());
            builder.setNumDistinctValues(longData.getNumDVs());
            if (longData.isSetBitVectors()) {
                builder.setBitVectors(longData.getBitVectors());
            }
            builder.setLongStats(HbaseMetastoreProto.ColumnStats.LongStats.newBuilder().setLowValue(longData.getLowValue()).setHighValue(longData.getHighValue()).build());
            break;
        case DOUBLE_STATS:
            DoubleColumnStatsData doubleData = colData.getDoubleStats();
            builder.setNumNulls(doubleData.getNumNulls());
            builder.setNumDistinctValues(doubleData.getNumDVs());
            if (doubleData.isSetBitVectors()) {
                builder.setBitVectors(doubleData.getBitVectors());
            }
            builder.setDoubleStats(HbaseMetastoreProto.ColumnStats.DoubleStats.newBuilder().setLowValue(doubleData.getLowValue()).setHighValue(doubleData.getHighValue()).build());
            break;
        case STRING_STATS:
            StringColumnStatsData stringData = colData.getStringStats();
            builder.setNumNulls(stringData.getNumNulls());
            builder.setNumDistinctValues(stringData.getNumDVs());
            if (stringData.isSetBitVectors()) {
                builder.setBitVectors(stringData.getBitVectors());
            }
            builder.setStringStats(HbaseMetastoreProto.ColumnStats.StringStats.newBuilder().setMaxColLength(stringData.getMaxColLen()).setAvgColLength(stringData.getAvgColLen()).build());
            break;
        case BINARY_STATS:
            BinaryColumnStatsData binaryData = colData.getBinaryStats();
            builder.setNumNulls(binaryData.getNumNulls());
            builder.setBinaryStats(HbaseMetastoreProto.ColumnStats.StringStats.newBuilder().setMaxColLength(binaryData.getMaxColLen()).setAvgColLength(binaryData.getAvgColLen()).build());
            break;
        case DECIMAL_STATS:
            DecimalColumnStatsData decimalData = colData.getDecimalStats();
            builder.setNumNulls(decimalData.getNumNulls());
            builder.setNumDistinctValues(decimalData.getNumDVs());
            if (decimalData.isSetBitVectors()) {
                builder.setBitVectors(decimalData.getBitVectors());
            }
            if (decimalData.getLowValue() != null && decimalData.getHighValue() != null) {
                builder.setDecimalStats(HbaseMetastoreProto.ColumnStats.DecimalStats.newBuilder().setLowValue(HbaseMetastoreProto.ColumnStats.DecimalStats.Decimal.newBuilder().setUnscaled(ByteString.copyFrom(decimalData.getLowValue().getUnscaled())).setScale(decimalData.getLowValue().getScale()).build()).setHighValue(HbaseMetastoreProto.ColumnStats.DecimalStats.Decimal.newBuilder().setUnscaled(ByteString.copyFrom(decimalData.getHighValue().getUnscaled())).setScale(decimalData.getHighValue().getScale()).build())).build();
            } else {
                builder.setDecimalStats(HbaseMetastoreProto.ColumnStats.DecimalStats.newBuilder().clear().build());
            }
            break;
        default:
            throw new RuntimeException("Woh, bad.  Unknown stats type!");
    }
    return builder.build();
}

Example 18 with BinaryColumnStatsData

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

the class MetaDataFormatUtils method formatWithIndentation.

private static void formatWithIndentation(String colName, String colType, String colComment, StringBuilder tableInfo, List<ColumnStatisticsObj> colStats) {
    tableInfo.append(String.format("%-" + ALIGNMENT + "s", colName)).append(FIELD_DELIM);
    tableInfo.append(String.format("%-" + ALIGNMENT + "s", colType)).append(FIELD_DELIM);
    if (colStats != null) {
        ColumnStatisticsObj cso = getColumnStatisticsObject(colName, colType, colStats);
        if (cso != null) {
            ColumnStatisticsData csd = cso.getStatsData();
            if (csd.isSetBinaryStats()) {
                BinaryColumnStatsData bcsd = csd.getBinaryStats();
                appendColumnStats(tableInfo, "", "", bcsd.getNumNulls(), "", bcsd.getAvgColLen(), bcsd.getMaxColLen(), "", "");
            } else if (csd.isSetStringStats()) {
                StringColumnStatsData scsd = csd.getStringStats();
                appendColumnStats(tableInfo, "", "", scsd.getNumNulls(), scsd.getNumDVs(), scsd.getAvgColLen(), scsd.getMaxColLen(), "", "");
            } else if (csd.isSetBooleanStats()) {
                BooleanColumnStatsData bcsd = csd.getBooleanStats();
                appendColumnStats(tableInfo, "", "", bcsd.getNumNulls(), "", "", "", bcsd.getNumTrues(), bcsd.getNumFalses());
            } else if (csd.isSetDecimalStats()) {
                DecimalColumnStatsData dcsd = csd.getDecimalStats();
                appendColumnStats(tableInfo, convertToString(dcsd.getLowValue()), convertToString(dcsd.getHighValue()), dcsd.getNumNulls(), dcsd.getNumDVs(), "", "", "", "");
            } else if (csd.isSetDoubleStats()) {
                DoubleColumnStatsData dcsd = csd.getDoubleStats();
                appendColumnStats(tableInfo, dcsd.getLowValue(), dcsd.getHighValue(), dcsd.getNumNulls(), dcsd.getNumDVs(), "", "", "", "");
            } else if (csd.isSetLongStats()) {
                LongColumnStatsData lcsd = csd.getLongStats();
                appendColumnStats(tableInfo, lcsd.getLowValue(), lcsd.getHighValue(), lcsd.getNumNulls(), lcsd.getNumDVs(), "", "", "", "");
            } else if (csd.isSetDateStats()) {
                DateColumnStatsData dcsd = csd.getDateStats();
                appendColumnStats(tableInfo, convertToString(dcsd.getLowValue()), convertToString(dcsd.getHighValue()), dcsd.getNumNulls(), dcsd.getNumDVs(), "", "", "", "");
            }
        } else {
            appendColumnStats(tableInfo, "", "", "", "", "", "", "", "");
        }
    }
    int colNameLength = ALIGNMENT > colName.length() ? ALIGNMENT : colName.length();
    int colTypeLength = ALIGNMENT > colType.length() ? ALIGNMENT : colType.length();
    indentMultilineValue(colComment, tableInfo, new int[] { colNameLength, colTypeLength }, false);
}

Example 19 with BinaryColumnStatsData

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

the class MetaDataFormatUtils method formatWithoutIndentation.

private static void formatWithoutIndentation(String name, String type, String comment, StringBuilder colBuffer, List<ColumnStatisticsObj> colStats) {
    colBuffer.append(name);
    colBuffer.append(FIELD_DELIM);
    colBuffer.append(type);
    colBuffer.append(FIELD_DELIM);
    if (colStats != null) {
        ColumnStatisticsObj cso = getColumnStatisticsObject(name, type, colStats);
        if (cso != null) {
            ColumnStatisticsData csd = cso.getStatsData();
            if (csd.isSetBinaryStats()) {
                BinaryColumnStatsData bcsd = csd.getBinaryStats();
                appendColumnStatsNoFormatting(colBuffer, "", "", bcsd.getNumNulls(), "", bcsd.getAvgColLen(), bcsd.getMaxColLen(), "", "");
            } else if (csd.isSetStringStats()) {
                StringColumnStatsData scsd = csd.getStringStats();
                appendColumnStatsNoFormatting(colBuffer, "", "", scsd.getNumNulls(), scsd.getNumDVs(), scsd.getAvgColLen(), scsd.getMaxColLen(), "", "");
            } else if (csd.isSetBooleanStats()) {
                BooleanColumnStatsData bcsd = csd.getBooleanStats();
                appendColumnStatsNoFormatting(colBuffer, "", "", bcsd.getNumNulls(), "", "", "", bcsd.getNumTrues(), bcsd.getNumFalses());
            } else if (csd.isSetDecimalStats()) {
                DecimalColumnStatsData dcsd = csd.getDecimalStats();
                appendColumnStatsNoFormatting(colBuffer, convertToString(dcsd.getLowValue()), convertToString(dcsd.getHighValue()), dcsd.getNumNulls(), dcsd.getNumDVs(), "", "", "", "");
            } else if (csd.isSetDoubleStats()) {
                DoubleColumnStatsData dcsd = csd.getDoubleStats();
                appendColumnStatsNoFormatting(colBuffer, dcsd.getLowValue(), dcsd.getHighValue(), dcsd.getNumNulls(), dcsd.getNumDVs(), "", "", "", "");
            } else if (csd.isSetLongStats()) {
                LongColumnStatsData lcsd = csd.getLongStats();
                appendColumnStatsNoFormatting(colBuffer, lcsd.getLowValue(), lcsd.getHighValue(), lcsd.getNumNulls(), lcsd.getNumDVs(), "", "", "", "");
            } else if (csd.isSetDateStats()) {
                DateColumnStatsData dcsd = csd.getDateStats();
                appendColumnStatsNoFormatting(colBuffer, convertToString(dcsd.getLowValue()), convertToString(dcsd.getHighValue()), dcsd.getNumNulls(), dcsd.getNumDVs(), "", "", "", "");
            }
        } else {
            appendColumnStatsNoFormatting(colBuffer, "", "", "", "", "", "", "", "");
        }
    }
    colBuffer.append(comment == null ? "" : HiveStringUtils.escapeJava(comment));
    colBuffer.append(LINE_DELIM);
}

Example 20 with BinaryColumnStatsData

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

the class StatObjectConverter method fillColumnStatisticsData.

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) 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")) {
        StringColumnStatsData stringStats = new StringColumnStatsData();
        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")) {
        LongColumnStatsData longStats = new LongColumnStatsData();
        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);
        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.
            long estimation = MetaStoreDirectSql.extractSqlLong((MetaStoreDirectSql.extractSqlLong(lhigh) - MetaStoreDirectSql.extractSqlLong(llow)) / MetaStoreDirectSql.extractSqlDouble(avgLong));
            if (estimation < lowerBound) {
                longStats.setNumDVs(lowerBound);
            } else if (estimation > higherBound) {
                longStats.setNumDVs(higherBound);
            } else {
                longStats.setNumDVs(estimation);
            }
        } else {
            longStats.setNumDVs(lowerBound);
        }
        data.setLongStats(longStats);
    } else if (colType.equals("date")) {
        DateColumnStatsData dateStats = new DateColumnStatsData();
        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);
        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.
            long estimation = MetaStoreDirectSql.extractSqlLong((MetaStoreDirectSql.extractSqlLong(lhigh) - MetaStoreDirectSql.extractSqlLong(llow)) / MetaStoreDirectSql.extractSqlDouble(avgLong));
            if (estimation < lowerBound) {
                dateStats.setNumDVs(lowerBound);
            } else if (estimation > higherBound) {
                dateStats.setNumDVs(higherBound);
            } else {
                dateStats.setNumDVs(estimation);
            }
        } else {
            dateStats.setNumDVs(lowerBound);
        }
        data.setDateStats(dateStats);
    } else if (colType.equals("double") || colType.equals("float")) {
        DoubleColumnStatsData doubleStats = new DoubleColumnStatsData();
        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(lowerBound);
        }
        data.setDoubleStats(doubleStats);
    } else if (colType.startsWith("decimal")) {
        DecimalColumnStatsData decimalStats = new DecimalColumnStatsData();
        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(lowerBound);
        }
        data.setDecimalStats(decimalStats);
    }
}