Examples with ColumnStatistics io.prestosql.spi.statistics.ColumnStatistics used on opensource projects

Example 6 with ColumnStatistics

use of io.prestosql.spi.statistics.ColumnStatistics in project hetu-core by openlookeng.

the class AbstractTestHive method assertTableStatsComputed.

private void assertTableStatsComputed(SchemaTableName tableName, Set<String> expectedColumnStatsColumns) {
    try (Transaction transaction = newTransaction()) {
        ConnectorMetadata metadata = transaction.getMetadata();
        ConnectorSession session = newSession();
        ConnectorTableHandle tableHandle = getTableHandle(metadata, tableName);
        TableStatistics tableStatistics = metadata.getTableStatistics(session, tableHandle, Constraint.alwaysTrue(), true);
        assertFalse(tableStatistics.getRowCount().isUnknown(), "row count is unknown");
        Map<String, ColumnStatistics> columnsStatistics = tableStatistics.getColumnStatistics().entrySet().stream().collect(toImmutableMap(entry -> ((HiveColumnHandle) entry.getKey()).getName(), Map.Entry::getValue));
        assertEquals(columnsStatistics.keySet(), expectedColumnStatsColumns, "columns with statistics");
        Map<String, ColumnHandle> columnHandles = metadata.getColumnHandles(session, tableHandle);
        columnsStatistics.forEach((columnName, columnStatistics) -> {
            ColumnHandle columnHandle = columnHandles.get(columnName);
            Type columnType = metadata.getColumnMetadata(session, tableHandle, columnHandle).getType();
            assertFalse(columnStatistics.getNullsFraction().isUnknown(), "unknown nulls fraction for " + columnName);
            assertFalse(columnStatistics.getDistinctValuesCount().isUnknown(), "unknown distinct values count for " + columnName);
            if (isVarcharType(columnType)) {
                assertFalse(columnStatistics.getDataSize().isUnknown(), "unknown data size for " + columnName);
            } else {
                assertTrue(columnStatistics.getDataSize().isUnknown(), "unknown data size for" + columnName);
            }
        });
    }
}

Example 7 with ColumnStatistics

use of io.prestosql.spi.statistics.ColumnStatistics in project hetu-core by openlookeng.

the class MetastoreHiveStatisticsProvider method getTableStatistics.

private static TableStatistics getTableStatistics(Map<String, ColumnHandle> columns, Map<String, Type> columnTypes, List<HivePartition> partitions, Map<String, PartitionStatistics> statistics) {
    if (statistics.isEmpty()) {
        return TableStatistics.empty();
    }
    checkArgument(!partitions.isEmpty(), "partitions is empty");
    OptionalDouble optionalAverageRowsPerPartition = calculateAverageRowsPerPartition(statistics.values());
    if (!optionalAverageRowsPerPartition.isPresent()) {
        return TableStatistics.empty();
    }
    double averageRowsPerPartition = optionalAverageRowsPerPartition.getAsDouble();
    verify(averageRowsPerPartition >= 0, "averageRowsPerPartition must be greater than or equal to zero");
    int queriedPartitionsCount = partitions.size();
    double rowCount = averageRowsPerPartition * queriedPartitionsCount;
    TableStatistics.Builder result = TableStatistics.builder();
    long fileCount = calulateFileCount(statistics.values());
    long totalOnDiskSize = calculateTotalOnDiskSizeInBytes(statistics.values());
    result.setRowCount(Estimate.of(rowCount));
    result.setFileCount(fileCount);
    result.setOnDiskDataSizeInBytes(totalOnDiskSize);
    for (Map.Entry<String, ColumnHandle> column : columns.entrySet()) {
        String columnName = column.getKey();
        HiveColumnHandle columnHandle = (HiveColumnHandle) column.getValue();
        Type columnType = columnTypes.get(columnName);
        ColumnStatistics columnStatistics;
        TableColumnStatistics tableColumnStatistics;
        if (columnHandle.isPartitionKey()) {
            tableColumnStatistics = statsCache.get(partitions.get(0).getTableName().getTableName() + columnName);
            if (tableColumnStatistics == null || invalidateStatsCache(tableColumnStatistics, Estimate.of(rowCount), fileCount, totalOnDiskSize)) {
                columnStatistics = createPartitionColumnStatistics(columnHandle, columnType, partitions, statistics, averageRowsPerPartition, rowCount);
                TableStatistics tableStatistics = new TableStatistics(Estimate.of(rowCount), fileCount, totalOnDiskSize, ImmutableMap.of());
                tableColumnStatistics = new TableColumnStatistics(tableStatistics, columnStatistics);
                statsCache.put(partitions.get(0).getTableName().getTableName() + columnName, tableColumnStatistics);
            } else {
                columnStatistics = tableColumnStatistics.columnStatistics;
            }
        } else {
            columnStatistics = createDataColumnStatistics(columnName, columnType, rowCount, statistics.values());
        }
        result.setColumnStatistics(columnHandle, columnStatistics);
    }
    return result.build();
}

Example 8 with ColumnStatistics

use of io.prestosql.spi.statistics.ColumnStatistics in project boostkit-bigdata by kunpengcompute.

the class AbstractTestHive method assertTableStatsComputed.

private void assertTableStatsComputed(SchemaTableName tableName, Set<String> expectedColumnStatsColumns) {
    try (Transaction transaction = newTransaction()) {
        ConnectorMetadata metadata = transaction.getMetadata();
        ConnectorSession session = newSession();
        ConnectorTableHandle tableHandle = getTableHandle(metadata, tableName);
        TableStatistics tableStatistics = metadata.getTableStatistics(session, tableHandle, Constraint.alwaysTrue(), true);
        assertFalse(tableStatistics.getRowCount().isUnknown(), "row count is unknown");
        Map<String, ColumnStatistics> columnsStatistics = tableStatistics.getColumnStatistics().entrySet().stream().collect(toImmutableMap(entry -> ((HiveColumnHandle) entry.getKey()).getName(), Map.Entry::getValue));
        assertEquals(columnsStatistics.keySet(), expectedColumnStatsColumns, "columns with statistics");
        Map<String, ColumnHandle> columnHandles = metadata.getColumnHandles(session, tableHandle);
        columnsStatistics.forEach((columnName, columnStatistics) -> {
            ColumnHandle columnHandle = columnHandles.get(columnName);
            Type columnType = metadata.getColumnMetadata(session, tableHandle, columnHandle).getType();
            assertFalse(columnStatistics.getNullsFraction().isUnknown(), "unknown nulls fraction for " + columnName);
            assertFalse(columnStatistics.getDistinctValuesCount().isUnknown(), "unknown distinct values count for " + columnName);
            if (isVarcharType(columnType)) {
                assertFalse(columnStatistics.getDataSize().isUnknown(), "unknown data size for " + columnName);
            } else {
                assertTrue(columnStatistics.getDataSize().isUnknown(), "unknown data size for" + columnName);
            }
        });
    }
}

Example 9 with ColumnStatistics

use of io.prestosql.spi.statistics.ColumnStatistics in project boostkit-bigdata by kunpengcompute.

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 10 with ColumnStatistics

use of io.prestosql.spi.statistics.ColumnStatistics in project boostkit-bigdata by kunpengcompute.

the class MetastoreHiveStatisticsProvider method createZeroStatistics.

private TableStatistics createZeroStatistics(Map<String, ColumnHandle> columns, Map<String, Type> columnTypes) {
    TableStatistics.Builder result = TableStatistics.builder();
    result.setRowCount(Estimate.of(0));
    columns.forEach((columnName, columnHandle) -> {
        Type columnType = columnTypes.get(columnName);
        verify(columnType != null, "columnType is missing for column: %s", columnName);
        ColumnStatistics.Builder columnStatistics = ColumnStatistics.builder();
        columnStatistics.setNullsFraction(Estimate.of(0));
        columnStatistics.setDistinctValuesCount(Estimate.of(0));
        if (hasDataSize(columnType)) {
            columnStatistics.setDataSize(Estimate.of(0));
        }
        result.setColumnStatistics(columnHandle, columnStatistics.build());
    });
    return result.build();
}