Examples with HiveColumnStatistics com.facebook.presto.hive.metastore.HiveColumnStatistics used on opensource projects

Example 21 with HiveColumnStatistics

use of com.facebook.presto.hive.metastore.HiveColumnStatistics in project presto by prestodb.

the class FileHiveMetastore method getTableStatistics.

@Override
public synchronized PartitionStatistics getTableStatistics(MetastoreContext metastoreContext, String databaseName, String tableName) {
    Path tableMetadataDirectory = getTableMetadataDirectory(databaseName, tableName);
    TableMetadata tableMetadata = readSchemaFile("table", tableMetadataDirectory, tableCodec).orElseThrow(() -> new TableNotFoundException(new SchemaTableName(databaseName, tableName)));
    HiveBasicStatistics basicStatistics = getHiveBasicStatistics(tableMetadata.getParameters());
    Map<String, HiveColumnStatistics> columnStatistics = tableMetadata.getColumnStatistics();
    return new PartitionStatistics(basicStatistics, columnStatistics);
}

Example 22 with HiveColumnStatistics

use of com.facebook.presto.hive.metastore.HiveColumnStatistics in project presto by prestodb.

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().isPresent()) {
            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);
        verify(columnStatistics != null, "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);
}

Example 23 with HiveColumnStatistics

use of com.facebook.presto.hive.metastore.HiveColumnStatistics in project presto by prestodb.

the class MetastoreHiveStatisticsProvider method calculateNullsFraction.

@VisibleForTesting
static Estimate calculateNullsFraction(String column, Collection<PartitionStatistics> partitionStatistics) {
    List<PartitionStatistics> statisticsWithKnownRowCountAndNullsCount = partitionStatistics.stream().filter(statistics -> {
        if (!statistics.getBasicStatistics().getRowCount().isPresent()) {
            return false;
        }
        HiveColumnStatistics columnStatistics = statistics.getColumnStatistics().get(column);
        if (columnStatistics == null) {
            return false;
        }
        return columnStatistics.getNullsCount().isPresent();
    }).collect(toImmutableList());
    if (statisticsWithKnownRowCountAndNullsCount.isEmpty()) {
        return Estimate.unknown();
    }
    long totalNullsCount = 0;
    long totalRowCount = 0;
    for (PartitionStatistics statistics : statisticsWithKnownRowCountAndNullsCount) {
        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);
        verify(columnStatistics != null, "columnStatistics is null");
        long nullsCount = columnStatistics.getNullsCount().orElseThrow(() -> new VerifyException("nullsCount is not present"));
        verify(nullsCount >= 0, "nullsCount must be greater than or equal to zero");
        verify(nullsCount <= rowCount, "nullsCount must be less than or equal to rowCount. nullsCount: %s. rowCount: %s.", nullsCount, rowCount);
        totalNullsCount += nullsCount;
        totalRowCount += rowCount;
    }
    if (totalRowCount == 0) {
        return Estimate.zero();
    }
    verify(totalNullsCount <= totalRowCount, "totalNullsCount must be less than or equal to totalRowCount. totalNullsCount: %s. totalRowCount: %s.", totalNullsCount, totalRowCount);
    return Estimate.of(((double) totalNullsCount) / totalRowCount);
}