result.setNullsCount(rowCount - BIGINT.getLong(computedStatistics.get(NUMBER_OF_NON_NULL_VALUES), 0));
private static PartitionStatistics nullsCount(long nullsCount) { return new PartitionStatistics(HiveBasicStatistics.createEmptyStatistics(), ImmutableMap.of(COLUMN, HiveColumnStatistics.builder().setNullsCount(nullsCount).build())); }
private static PartitionStatistics rowsCountAndNullsCount(long rowsCount, long nullsCount) { return new PartitionStatistics( new HiveBasicStatistics(0, rowsCount, 0, 0), ImmutableMap.of(COLUMN, HiveColumnStatistics.builder().setNullsCount(nullsCount).build())); }
PartitionStatistics.builder() .setBasicStatistics(new HiveBasicStatistics(0, 0, 0, 0)) .setColumnStatistics(ImmutableMap.of(COLUMN, HiveColumnStatistics.builder().setNullsCount(-1).build())) .build(), invalidColumnStatistics("nullsCount must be greater than or equal to zero: -1")); PartitionStatistics.builder() .setBasicStatistics(new HiveBasicStatistics(0, 0, 0, 0)) .setColumnStatistics(ImmutableMap.of(COLUMN, HiveColumnStatistics.builder().setNullsCount(1).build())) .build(), invalidColumnStatistics("nullsCount must be less than or equal to rowCount. nullsCount: 1. rowCount: 0.")); PartitionStatistics.builder() .setBasicStatistics(new HiveBasicStatistics(0, 1, 0, 0)) .setColumnStatistics(ImmutableMap.of(COLUMN, HiveColumnStatistics.builder().setDistinctValuesCount(1).setNullsCount(1).build())) .build(), invalidColumnStatistics("distinctValuesCount must be less than or equal to nonNullsCount. distinctValuesCount: 1. nonNullsCount: 0."));
@Test public void testMergeGenericColumnStatistics() { assertMergeHiveColumnStatistics( HiveColumnStatistics.builder().setDistinctValuesCount(OptionalLong.empty()).build(), HiveColumnStatistics.builder().setDistinctValuesCount(OptionalLong.empty()).build(), HiveColumnStatistics.builder().setDistinctValuesCount(OptionalLong.empty()).build()); assertMergeHiveColumnStatistics( HiveColumnStatistics.builder().setDistinctValuesCount(OptionalLong.of(1)).build(), HiveColumnStatistics.builder().setDistinctValuesCount(OptionalLong.empty()).build(), HiveColumnStatistics.builder().setDistinctValuesCount(OptionalLong.empty()).build()); assertMergeHiveColumnStatistics( HiveColumnStatistics.builder().setDistinctValuesCount(OptionalLong.of(1)).build(), HiveColumnStatistics.builder().setDistinctValuesCount(OptionalLong.of(2)).build(), HiveColumnStatistics.builder().setDistinctValuesCount(OptionalLong.of(2)).build()); assertMergeHiveColumnStatistics( HiveColumnStatistics.builder().setNullsCount(OptionalLong.empty()).build(), HiveColumnStatistics.builder().setNullsCount(OptionalLong.empty()).build(), HiveColumnStatistics.builder().setNullsCount(OptionalLong.empty()).build()); assertMergeHiveColumnStatistics( HiveColumnStatistics.builder().setNullsCount(OptionalLong.of(1)).build(), HiveColumnStatistics.builder().setNullsCount(OptionalLong.empty()).build(), HiveColumnStatistics.builder().setNullsCount(OptionalLong.empty()).build()); assertMergeHiveColumnStatistics( HiveColumnStatistics.builder().setNullsCount(OptionalLong.of(1)).build(), HiveColumnStatistics.builder().setNullsCount(OptionalLong.of(2)).build(), HiveColumnStatistics.builder().setNullsCount(OptionalLong.of(3)).build()); }
private static PartitionStatistics nullsCount(long nullsCount) { return new PartitionStatistics(HiveBasicStatistics.createEmptyStatistics(), ImmutableMap.of(COLUMN, HiveColumnStatistics.builder().setNullsCount(nullsCount).build())); }
private static PartitionStatistics rowsCountAndNullsCount(long rowsCount, long nullsCount) { return new PartitionStatistics( new HiveBasicStatistics(0, rowsCount, 0, 0), ImmutableMap.of(COLUMN, HiveColumnStatistics.builder().setNullsCount(nullsCount).build())); }
PartitionStatistics.builder() .setBasicStatistics(new HiveBasicStatistics(0, 0, 0, 0)) .setColumnStatistics(ImmutableMap.of(COLUMN, HiveColumnStatistics.builder().setNullsCount(-1).build())) .build(), invalidColumnStatistics("nullsCount must be greater than or equal to zero: -1")); PartitionStatistics.builder() .setBasicStatistics(new HiveBasicStatistics(0, 0, 0, 0)) .setColumnStatistics(ImmutableMap.of(COLUMN, HiveColumnStatistics.builder().setNullsCount(1).build())) .build(), invalidColumnStatistics("nullsCount must be less than or equal to rowCount. nullsCount: 1. rowCount: 0.")); PartitionStatistics.builder() .setBasicStatistics(new HiveBasicStatistics(0, 1, 0, 0)) .setColumnStatistics(ImmutableMap.of(COLUMN, HiveColumnStatistics.builder().setDistinctValuesCount(1).setNullsCount(1).build())) .build(), invalidColumnStatistics("distinctValuesCount must be less than or equal to nonNullsCount. distinctValuesCount: 1. nonNullsCount: 0."));
@Test public void testMergeGenericColumnStatistics() { assertMergeHiveColumnStatistics( HiveColumnStatistics.builder().setDistinctValuesCount(OptionalLong.empty()).build(), HiveColumnStatistics.builder().setDistinctValuesCount(OptionalLong.empty()).build(), HiveColumnStatistics.builder().setDistinctValuesCount(OptionalLong.empty()).build()); assertMergeHiveColumnStatistics( HiveColumnStatistics.builder().setDistinctValuesCount(OptionalLong.of(1)).build(), HiveColumnStatistics.builder().setDistinctValuesCount(OptionalLong.empty()).build(), HiveColumnStatistics.builder().setDistinctValuesCount(OptionalLong.empty()).build()); assertMergeHiveColumnStatistics( HiveColumnStatistics.builder().setDistinctValuesCount(OptionalLong.of(1)).build(), HiveColumnStatistics.builder().setDistinctValuesCount(OptionalLong.of(2)).build(), HiveColumnStatistics.builder().setDistinctValuesCount(OptionalLong.of(2)).build()); assertMergeHiveColumnStatistics( HiveColumnStatistics.builder().setNullsCount(OptionalLong.empty()).build(), HiveColumnStatistics.builder().setNullsCount(OptionalLong.empty()).build(), HiveColumnStatistics.builder().setNullsCount(OptionalLong.empty()).build()); assertMergeHiveColumnStatistics( HiveColumnStatistics.builder().setNullsCount(OptionalLong.of(1)).build(), HiveColumnStatistics.builder().setNullsCount(OptionalLong.empty()).build(), HiveColumnStatistics.builder().setNullsCount(OptionalLong.empty()).build()); assertMergeHiveColumnStatistics( HiveColumnStatistics.builder().setNullsCount(OptionalLong.of(1)).build(), HiveColumnStatistics.builder().setNullsCount(OptionalLong.of(2)).build(), HiveColumnStatistics.builder().setNullsCount(OptionalLong.of(3)).build()); }
public static HiveColumnStatistics createIntegerColumnStatistics(OptionalLong min, OptionalLong max, OptionalLong nullsCount, OptionalLong distinctValuesCount) { return builder() .setIntegerStatistics(new IntegerStatistics(min, max)) .setNullsCount(nullsCount) .setDistinctValuesCount(distinctValuesCount) .build(); }
public static HiveColumnStatistics createStringColumnStatistics( OptionalLong maxValueSizeInBytes, OptionalLong totalSizeInBytes, OptionalLong nullsCount, OptionalLong distinctValuesCount) { return builder() .setMaxValueSizeInBytes(maxValueSizeInBytes) .setTotalSizeInBytes(totalSizeInBytes) .setNullsCount(nullsCount) .setDistinctValuesCount(distinctValuesCount) .build(); }
public static HiveColumnStatistics createBinaryColumnStatistics(OptionalLong maxValueSizeInBytes, OptionalLong totalSizeInBytes, OptionalLong nullsCount) { return builder() .setMaxValueSizeInBytes(maxValueSizeInBytes) .setTotalSizeInBytes(totalSizeInBytes) .setNullsCount(nullsCount) .build(); }
public static HiveColumnStatistics createDateColumnStatistics(Optional<LocalDate> min, Optional<LocalDate> max, OptionalLong nullsCount, OptionalLong distinctValuesCount) { return builder() .setDateStatistics(new DateStatistics(min, max)) .setNullsCount(nullsCount) .setDistinctValuesCount(distinctValuesCount) .build(); }
public static HiveColumnStatistics createDecimalColumnStatistics(Optional<BigDecimal> min, Optional<BigDecimal> max, OptionalLong nullsCount, OptionalLong distinctValuesCount) { return builder() .setDecimalStatistics(new DecimalStatistics(min, max)) .setNullsCount(nullsCount) .setDistinctValuesCount(distinctValuesCount) .build(); }
public static HiveColumnStatistics createDoubleColumnStatistics(OptionalDouble min, OptionalDouble max, OptionalLong nullsCount, OptionalLong distinctValuesCount) { return builder() .setDoubleStatistics(new DoubleStatistics(min, max)) .setNullsCount(nullsCount) .setDistinctValuesCount(distinctValuesCount) .build(); }
public static HiveColumnStatistics createBooleanColumnStatistics(OptionalLong trueCount, OptionalLong falseCount, OptionalLong nullsCount) { return builder() .setBooleanStatistics(new BooleanStatistics(trueCount, falseCount)) .setNullsCount(nullsCount) .build(); }