Examples with HiveColumnStatistics io.trino.plugin.hive.metastore.HiveColumnStatistics used on opensource projects

Example 1 with HiveColumnStatistics

use of io.trino.plugin.hive.metastore.HiveColumnStatistics in project trino by trinodb.

the class FileHiveMetastore method getTableStatistics.

private synchronized PartitionStatistics getTableStatistics(String databaseName, String tableName) {
    Path tableMetadataDirectory = getTableMetadataDirectory(databaseName, tableName);
    TableMetadata tableMetadata = readSchemaFile(TABLE, tableMetadataDirectory, tableCodec).orElseThrow(() -> new TableNotFoundException(new SchemaTableName(databaseName, tableName)));
    checkVersion(tableMetadata.getWriterVersion());
    HiveBasicStatistics basicStatistics = getHiveBasicStatistics(tableMetadata.getParameters());
    Map<String, HiveColumnStatistics> columnStatistics = tableMetadata.getColumnStatistics();
    return new PartitionStatistics(basicStatistics, columnStatistics);
}

Example 2 with HiveColumnStatistics

use of io.trino.plugin.hive.metastore.HiveColumnStatistics in project trino by trinodb.

the class GlueStatConverter method toColumnStatistics.

private static ColumnStatistics toColumnStatistics(Column column, HiveColumnStatistics statistics, OptionalLong rowCount) {
    ColumnStatistics columnStatistics = new ColumnStatistics();
    HiveType columnType = column.getType();
    columnStatistics.setColumnName(column.getName());
    columnStatistics.setColumnType(columnType.toString());
    ColumnStatisticsData catalogColumnStatisticsData = toGlueColumnStatisticsData(statistics, columnType, rowCount);
    columnStatistics.setStatisticsData(catalogColumnStatisticsData);
    columnStatistics.setAnalyzedTime(new Date());
    return columnStatistics;
}

Example 3 with HiveColumnStatistics

use of io.trino.plugin.hive.metastore.HiveColumnStatistics in project trino by trinodb.

the class GlueStatConverter method toGlueColumnStatisticsData.

private static ColumnStatisticsData toGlueColumnStatisticsData(HiveColumnStatistics statistics, HiveType columnType, OptionalLong rowCount) {
    TypeInfo typeInfo = columnType.getTypeInfo();
    checkArgument(typeInfo.getCategory() == PRIMITIVE, "Unsupported statistics type: %s", columnType);
    ColumnStatisticsData catalogColumnStatisticsData = new ColumnStatisticsData();
    switch(((PrimitiveTypeInfo) typeInfo).getPrimitiveCategory()) {
        case BOOLEAN:
            {
                BooleanColumnStatisticsData data = new BooleanColumnStatisticsData();
                statistics.getNullsCount().ifPresent(data::setNumberOfNulls);
                statistics.getBooleanStatistics().ifPresent(booleanStatistics -> {
                    booleanStatistics.getFalseCount().ifPresent(data::setNumberOfFalses);
                    booleanStatistics.getTrueCount().ifPresent(data::setNumberOfTrues);
                });
                catalogColumnStatisticsData.setType(ColumnStatisticsType.BOOLEAN.toString());
                catalogColumnStatisticsData.setBooleanColumnStatisticsData(data);
                break;
            }
        case BINARY:
            {
                BinaryColumnStatisticsData data = new BinaryColumnStatisticsData();
                statistics.getNullsCount().ifPresent(data::setNumberOfNulls);
                data.setMaximumLength(statistics.getMaxValueSizeInBytes().orElse(0));
                data.setAverageLength(getAverageColumnLength(statistics.getTotalSizeInBytes(), rowCount, statistics.getNullsCount()).orElse(0));
                catalogColumnStatisticsData.setType(ColumnStatisticsType.BINARY.toString());
                catalogColumnStatisticsData.setBinaryColumnStatisticsData(data);
                break;
            }
        case DATE:
            {
                DateColumnStatisticsData data = new DateColumnStatisticsData();
                statistics.getDateStatistics().ifPresent(dateStatistics -> {
                    dateStatistics.getMin().ifPresent(value -> data.setMinimumValue(localDateToDate(value)));
                    dateStatistics.getMax().ifPresent(value -> data.setMaximumValue(localDateToDate(value)));
                });
                statistics.getNullsCount().ifPresent(data::setNumberOfNulls);
                toMetastoreDistinctValuesCount(statistics.getDistinctValuesCount(), statistics.getNullsCount()).ifPresent(data::setNumberOfDistinctValues);
                catalogColumnStatisticsData.setType(ColumnStatisticsType.DATE.toString());
                catalogColumnStatisticsData.setDateColumnStatisticsData(data);
                break;
            }
        case DECIMAL:
            {
                DecimalColumnStatisticsData data = new DecimalColumnStatisticsData();
                statistics.getDecimalStatistics().ifPresent(decimalStatistics -> {
                    decimalStatistics.getMin().ifPresent(value -> data.setMinimumValue(bigDecimalToGlueDecimal(value)));
                    decimalStatistics.getMax().ifPresent(value -> data.setMaximumValue(bigDecimalToGlueDecimal(value)));
                });
                statistics.getNullsCount().ifPresent(data::setNumberOfNulls);
                toMetastoreDistinctValuesCount(statistics.getDistinctValuesCount(), statistics.getNullsCount()).ifPresent(data::setNumberOfDistinctValues);
                catalogColumnStatisticsData.setType(ColumnStatisticsType.DECIMAL.toString());
                catalogColumnStatisticsData.setDecimalColumnStatisticsData(data);
                break;
            }
        case FLOAT:
        case DOUBLE:
            {
                DoubleColumnStatisticsData data = new DoubleColumnStatisticsData();
                statistics.getDoubleStatistics().ifPresent(doubleStatistics -> {
                    doubleStatistics.getMin().ifPresent(data::setMinimumValue);
                    doubleStatistics.getMax().ifPresent(data::setMaximumValue);
                });
                statistics.getNullsCount().ifPresent(data::setNumberOfNulls);
                toMetastoreDistinctValuesCount(statistics.getDistinctValuesCount(), statistics.getNullsCount()).ifPresent(data::setNumberOfDistinctValues);
                catalogColumnStatisticsData.setType(ColumnStatisticsType.DOUBLE.toString());
                catalogColumnStatisticsData.setDoubleColumnStatisticsData(data);
                break;
            }
        case BYTE:
        case SHORT:
        case INT:
        case LONG:
        case TIMESTAMP:
            {
                LongColumnStatisticsData data = new LongColumnStatisticsData();
                statistics.getIntegerStatistics().ifPresent(stats -> {
                    stats.getMin().ifPresent(data::setMinimumValue);
                    stats.getMax().ifPresent(data::setMaximumValue);
                });
                statistics.getNullsCount().ifPresent(data::setNumberOfNulls);
                toMetastoreDistinctValuesCount(statistics.getDistinctValuesCount(), statistics.getNullsCount()).ifPresent(data::setNumberOfDistinctValues);
                catalogColumnStatisticsData.setType(ColumnStatisticsType.LONG.toString());
                catalogColumnStatisticsData.setLongColumnStatisticsData(data);
                break;
            }
        case VARCHAR:
        case CHAR:
        case STRING:
            {
                StringColumnStatisticsData data = new StringColumnStatisticsData();
                statistics.getNullsCount().ifPresent(data::setNumberOfNulls);
                toMetastoreDistinctValuesCount(statistics.getDistinctValuesCount(), statistics.getNullsCount()).ifPresent(data::setNumberOfDistinctValues);
                data.setMaximumLength(statistics.getMaxValueSizeInBytes().orElse(0));
                data.setAverageLength(getAverageColumnLength(statistics.getTotalSizeInBytes(), rowCount, statistics.getNullsCount()).orElse(0));
                catalogColumnStatisticsData.setType(ColumnStatisticsType.STRING.toString());
                catalogColumnStatisticsData.setStringColumnStatisticsData(data);
                break;
            }
        default:
            throw new TrinoException(HIVE_INVALID_METADATA, "Invalid column statistics type: " + ((PrimitiveTypeInfo) typeInfo).getPrimitiveCategory());
    }
    return catalogColumnStatisticsData;
}

Example 4 with HiveColumnStatistics

use of io.trino.plugin.hive.metastore.HiveColumnStatistics in project trino by trinodb.

the class ThriftHiveMetastore method getTableStatistics.

@Override
public PartitionStatistics getTableStatistics(HiveIdentity identity, Table table) {
    List<String> dataColumns = table.getSd().getCols().stream().map(FieldSchema::getName).collect(toImmutableList());
    HiveBasicStatistics basicStatistics = getHiveBasicStatistics(table.getParameters());
    Map<String, HiveColumnStatistics> columnStatistics = getTableColumnStatistics(identity, table.getDbName(), table.getTableName(), dataColumns, basicStatistics.getRowCount());
    return new PartitionStatistics(basicStatistics, columnStatistics);
}

Example 5 with HiveColumnStatistics

use of io.trino.plugin.hive.metastore.HiveColumnStatistics in project trino by trinodb.

the class MetastoreHiveStatisticsProvider method calculateDataSize.

@VisibleForTesting
static Estimate calculateDataSize(String column, Collection<PartitionStatistics> partitionStatistics, double totalRowCount) {
    List<PartitionStatistics> statisticsWithKnownRowCountAndDataSize = partitionStatistics.stream().filter(statistics -> {
        if (statistics.getBasicStatistics().getRowCount().isEmpty()) {
            return false;
        }
        HiveColumnStatistics columnStatistics = statistics.getColumnStatistics().get(column);
        if (columnStatistics == null) {
            return false;
        }
        return columnStatistics.getTotalSizeInBytes().isPresent();
    }).collect(toImmutableList());
    if (statisticsWithKnownRowCountAndDataSize.isEmpty()) {
        return Estimate.unknown();
    }
    long knownRowCount = 0;
    long knownDataSize = 0;
    for (PartitionStatistics statistics : statisticsWithKnownRowCountAndDataSize) {
        long rowCount = statistics.getBasicStatistics().getRowCount().orElseThrow(() -> new VerifyException("rowCount is not present"));
        verify(rowCount >= 0, "rowCount must be greater than or equal to zero");
        HiveColumnStatistics columnStatistics = statistics.getColumnStatistics().get(column);
        verifyNotNull(columnStatistics, "columnStatistics is null");
        long dataSize = columnStatistics.getTotalSizeInBytes().orElseThrow(() -> new VerifyException("totalSizeInBytes is not present"));
        verify(dataSize >= 0, "dataSize must be greater than or equal to zero");
        knownRowCount += rowCount;
        knownDataSize += dataSize;
    }
    if (totalRowCount == 0) {
        return Estimate.zero();
    }
    if (knownRowCount == 0) {
        return Estimate.unknown();
    }
    double averageValueDataSizeInBytes = ((double) knownDataSize) / knownRowCount;
    return Estimate.of(averageValueDataSizeInBytes * totalRowCount);
}