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

Example 16 with ColumnStatistics

use of io.trino.spi.statistics.ColumnStatistics in project trino by trinodb.

the class TestTpchMetadata method testColumnStats.

private void testColumnStats(String schema, TpchTable<?> table, TpchColumn<?> column, Constraint constraint, ColumnStatistics expected) {
    TpchTableHandle tableHandle = tpchMetadata.getTableHandle(session, new SchemaTableName(schema, table.getTableName()));
    TableStatistics tableStatistics = tpchMetadata.getTableStatistics(session, tableHandle, constraint);
    ColumnHandle columnHandle = tpchMetadata.getColumnHandles(session, tableHandle).get(column.getSimplifiedColumnName());
    ColumnStatistics actual = tableStatistics.getColumnStatistics().get(columnHandle);
    EstimateAssertion estimateAssertion = new EstimateAssertion(TOLERANCE);
    estimateAssertion.assertClose(actual.getDistinctValuesCount(), expected.getDistinctValuesCount(), "distinctValuesCount");
    estimateAssertion.assertClose(actual.getDataSize(), expected.getDataSize(), "dataSize");
    estimateAssertion.assertClose(actual.getNullsFraction(), expected.getNullsFraction(), "nullsFraction");
    estimateAssertion.assertClose(actual.getRange(), expected.getRange(), "range");
}

Example 17 with ColumnStatistics

use of io.trino.spi.statistics.ColumnStatistics in project trino by trinodb.

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());
        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 (columnType instanceof VarcharType) {
                assertFalse(columnStatistics.getDataSize().isUnknown(), "unknown data size for " + columnName);
            } else {
                assertTrue(columnStatistics.getDataSize().isUnknown(), "unknown data size for" + columnName);
            }
        });
    }
}

Example 18 with ColumnStatistics

use of io.trino.spi.statistics.ColumnStatistics in project trino by trinodb.

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().isEmpty()) {
            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);
        verifyNotNull(columnStatistics, "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);
}

Example 19 with ColumnStatistics

use of io.trino.spi.statistics.ColumnStatistics in project trino by trinodb.

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");
    Optional<PartitionsRowCount> optionalRowCount = calculatePartitionsRowCount(statistics.values(), partitions.size());
    if (optionalRowCount.isEmpty()) {
        return TableStatistics.empty();
    }
    double rowCount = optionalRowCount.get().getRowCount();
    TableStatistics.Builder result = TableStatistics.builder();
    result.setRowCount(Estimate.of(rowCount));
    for (Map.Entry<String, ColumnHandle> column : columns.entrySet()) {
        String columnName = column.getKey();
        HiveColumnHandle columnHandle = (HiveColumnHandle) column.getValue();
        Type columnType = columnTypes.get(columnName);
        ColumnStatistics columnStatistics;
        if (columnHandle.isPartitionKey()) {
            double averageRowsPerPartition = optionalRowCount.get().getAverageRowsPerPartition();
            columnStatistics = createPartitionColumnStatistics(columnHandle, columnType, partitions, statistics, averageRowsPerPartition, rowCount);
        } else {
            columnStatistics = createDataColumnStatistics(columnName, columnType, rowCount, statistics.values());
        }
        result.setColumnStatistics(columnHandle, columnStatistics);
    }
    return result.build();
}

Example 20 with ColumnStatistics

use of io.trino.spi.statistics.ColumnStatistics in project trino by trinodb.

the class TestDeltaLakeMetastoreStatistics method testStatisticsParquetParsedStatisticsNullCount.

@Test
public void testStatisticsParquetParsedStatisticsNullCount() {
    // The transaction log for this table was created so that the checkpoints only write struct statistics, not json statistics
    // The table has one INTEGER column 'i' where 3 of the 9 values are null
    DeltaLakeTableHandle tableHandle = registerTable("parquet_struct_statistics_null_count");
    TableStatistics stats = deltaLakeMetastore.getTableStatistics(SESSION, tableHandle, Constraint.alwaysTrue());
    assertEquals(stats.getRowCount(), Estimate.of(9));
    Map<ColumnHandle, ColumnStatistics> statisticsMap = stats.getColumnStatistics();
    ColumnStatistics columnStats = statisticsMap.get(new DeltaLakeColumnHandle("i", INTEGER, REGULAR));
    assertEquals(columnStats.getNullsFraction(), Estimate.of(3.0 / 9.0));
}