Examples with ColumnStatistics com.facebook.presto.spi.statistics.ColumnStatistics used on opensource projects

Example 6 with ColumnStatistics

use of com.facebook.presto.spi.statistics.ColumnStatistics in project presto by prestodb.

the class TestConnectorFilterStatsCalculatorService method testTableStatisticsAfterFilter.

@Test
public void testTableStatisticsAfterFilter() {
    // totalSize always be zero
    assertPredicate("true", zeroTableStatistics, zeroTableStatistics);
    assertPredicate("x < 3e0", zeroTableStatistics, zeroTableStatistics);
    assertPredicate("false", zeroTableStatistics, zeroTableStatistics);
    // rowCount and totalSize all NaN
    assertPredicate("true", TableStatistics.empty(), TableStatistics.empty());
    // rowCount and totalSize from NaN to 0.0
    assertPredicate("false", TableStatistics.empty(), TableStatistics.builder().setRowCount(Estimate.zero()).setTotalSize(Estimate.zero()).build());
    TableStatistics filteredToZeroStatistics = TableStatistics.builder().setRowCount(Estimate.zero()).setTotalSize(Estimate.zero()).setColumnStatistics(xColumn, new ColumnStatistics(Estimate.of(1.0), Estimate.zero(), Estimate.zero(), Optional.empty())).build();
    assertPredicate("false", originalTableStatistics, filteredToZeroStatistics);
    TableStatistics filteredStatistics = TableStatistics.builder().setRowCount(Estimate.of(37.5)).setTotalSize(Estimate.of(300)).setColumnStatistics(xColumn, new ColumnStatistics(Estimate.zero(), Estimate.of(20), Estimate.unknown(), Optional.of(new DoubleRange(-10, 0)))).build();
    assertPredicate("x < 0", originalTableStatistics, filteredStatistics);
    TableStatistics filteredStatisticsWithoutTotalSize = TableStatistics.builder().setRowCount(Estimate.of(37.5)).setColumnStatistics(xColumn, new ColumnStatistics(Estimate.zero(), Estimate.of(20), Estimate.unknown(), Optional.of(new DoubleRange(-10, 0)))).build();
    assertPredicate("x < 0", originalTableStatisticsWithoutTotalSize, filteredStatisticsWithoutTotalSize);
}

Example 7 with ColumnStatistics

use of com.facebook.presto.spi.statistics.ColumnStatistics in project presto by prestodb.

the class AbstractTestHiveClient 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);
        List<ColumnHandle> allColumnHandles = ImmutableList.copyOf(metadata.getColumnHandles(session, tableHandle).values());
        TableStatistics tableStatistics = metadata.getTableStatistics(session, tableHandle, Optional.empty(), allColumnHandles, 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 (isVarcharType(columnType)) {
                assertFalse(columnStatistics.getDataSize().isUnknown(), "unknown data size for " + columnName);
            } else {
                assertTrue(columnStatistics.getDataSize().isUnknown(), "unknown data size for" + columnName);
            }
        });
    }
}

Example 8 with ColumnStatistics

use of com.facebook.presto.spi.statistics.ColumnStatistics 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 9 with ColumnStatistics

use of com.facebook.presto.spi.statistics.ColumnStatistics 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);
}

Example 10 with ColumnStatistics

use of com.facebook.presto.spi.statistics.ColumnStatistics in project presto by prestodb.

the class TestTpcdsMetadataStatistics method testTableStatisticsSerialization.

@Test
public void testTableStatisticsSerialization() {
    SchemaTableName schemaTableName = new SchemaTableName("sf1", Table.WEB_SITE.getName());
    ConnectorTableHandle tableHandle = metadata.getTableHandle(session, schemaTableName);
    List<ColumnHandle> columnHandles = ImmutableList.copyOf(metadata.getColumnHandles(session, tableHandle).values());
    TableStatistics tableStatistics = metadata.getTableStatistics(session, tableHandle, Optional.empty(), columnHandles, alwaysTrue());
    Entry<ColumnHandle, ColumnStatistics> entry = tableStatistics.getColumnStatistics().entrySet().iterator().next();
    TableStatistics expectedTableStatictics = tableStatistics.builder().setRowCount(tableStatistics.getRowCount()).setColumnStatistics(entry.getKey(), entry.getValue()).build();
    JsonCodec<TableStatistics> codec = JsonCodec.jsonCodec(TableStatistics.class);
    String json = codec.toJson(expectedTableStatictics);
    assertEquals(json, "{\n" + "  \"rowCount\" : {\n" + "    \"value\" : 30.0\n" + "  },\n" + "  \"totalSize\" : {\n" + "    \"value\" : \"NaN\"\n" + "  },\n" + "  \"columnStatistics\" : {\n" + "    \"tpcds:web_site_sk\" : {\n" + "      \"nullsFraction\" : {\n" + "        \"value\" : 0.0\n" + "      },\n" + "      \"distinctValuesCount\" : {\n" + "        \"value\" : 30.0\n" + "      },\n" + "      \"dataSize\" : {\n" + "        \"value\" : \"NaN\"\n" + "      },\n" + "      \"range\" : {\n" + "        \"min\" : 1.0,\n" + "        \"max\" : 30.0\n" + "      }\n" + "    }\n" + "  }\n" + "}");
}