Examples with DoubleStatistics io.trino.orc.metadata.statistics.DoubleStatistics used on opensource projects

Example 1 with DoubleStatistics

use of io.trino.orc.metadata.statistics.DoubleStatistics in project trino by trinodb.

the class IcebergOrcFileWriter method toIcebergMinMax.

private static Optional<IcebergMinMax> toIcebergMinMax(ColumnStatistics orcColumnStats, org.apache.iceberg.types.Type icebergType, MetricsModes.MetricsMode metricsModes) {
    BooleanStatistics booleanStatistics = orcColumnStats.getBooleanStatistics();
    if (booleanStatistics != null) {
        boolean hasTrueValues = booleanStatistics.getTrueValueCount() != 0;
        boolean hasFalseValues = orcColumnStats.getNumberOfValues() != booleanStatistics.getTrueValueCount();
        return Optional.of(new IcebergMinMax(icebergType, !hasFalseValues, hasTrueValues, metricsModes));
    }
    IntegerStatistics integerStatistics = orcColumnStats.getIntegerStatistics();
    if (integerStatistics != null) {
        Object min = integerStatistics.getMin();
        Object max = integerStatistics.getMax();
        if (min == null || max == null) {
            return Optional.empty();
        }
        if (icebergType.typeId() == org.apache.iceberg.types.Type.TypeID.INTEGER) {
            min = toIntExact((Long) min);
            max = toIntExact((Long) max);
        }
        return Optional.of(new IcebergMinMax(icebergType, min, max, metricsModes));
    }
    DoubleStatistics doubleStatistics = orcColumnStats.getDoubleStatistics();
    if (doubleStatistics != null) {
        Object min = doubleStatistics.getMin();
        Object max = doubleStatistics.getMax();
        if (min == null || max == null) {
            return Optional.empty();
        }
        if (icebergType.typeId() == org.apache.iceberg.types.Type.TypeID.FLOAT) {
            min = ((Double) min).floatValue();
            max = ((Double) max).floatValue();
        }
        return Optional.of(new IcebergMinMax(icebergType, min, max, metricsModes));
    }
    StringStatistics stringStatistics = orcColumnStats.getStringStatistics();
    if (stringStatistics != null) {
        Slice min = stringStatistics.getMin();
        Slice max = stringStatistics.getMax();
        if (min == null || max == null) {
            return Optional.empty();
        }
        return Optional.of(new IcebergMinMax(icebergType, min.toStringUtf8(), max.toStringUtf8(), metricsModes));
    }
    DateStatistics dateStatistics = orcColumnStats.getDateStatistics();
    if (dateStatistics != null) {
        Integer min = dateStatistics.getMin();
        Integer max = dateStatistics.getMax();
        if (min == null || max == null) {
            return Optional.empty();
        }
        return Optional.of(new IcebergMinMax(icebergType, min, max, metricsModes));
    }
    DecimalStatistics decimalStatistics = orcColumnStats.getDecimalStatistics();
    if (decimalStatistics != null) {
        BigDecimal min = decimalStatistics.getMin();
        BigDecimal max = decimalStatistics.getMax();
        if (min == null || max == null) {
            return Optional.empty();
        }
        min = min.setScale(((Types.DecimalType) icebergType).scale());
        max = max.setScale(((Types.DecimalType) icebergType).scale());
        return Optional.of(new IcebergMinMax(icebergType, min, max, metricsModes));
    }
    TimestampStatistics timestampStatistics = orcColumnStats.getTimestampStatistics();
    if (timestampStatistics != null) {
        Long min = timestampStatistics.getMin();
        Long max = timestampStatistics.getMax();
        if (min == null || max == null) {
            return Optional.empty();
        }
        // We are appending 999 microseconds to account for the fact that Trino ORC writer truncates timestamps.
        return Optional.of(new IcebergMinMax(icebergType, min * MICROSECONDS_PER_MILLISECOND, (max * MICROSECONDS_PER_MILLISECOND) + (MICROSECONDS_PER_MILLISECOND - 1), metricsModes));
    }
    return Optional.empty();
}