Examples with RichColumnDescriptor io.trino.parquet.RichColumnDescriptor used on opensource projects

Example 11 with RichColumnDescriptor

use of io.trino.parquet.RichColumnDescriptor in project trino by trinodb.

the class TestParquetPredicateUtils method testParquetTupleDomainMap.

@Test
public void testParquetTupleDomainMap() {
    MapType mapType = new MapType(INTEGER, INTEGER, new TypeOperators());
    HiveColumnHandle columnHandle = createBaseColumn("my_map", 0, HiveType.valueOf("map<int,int>"), mapType, REGULAR, Optional.empty());
    TupleDomain<HiveColumnHandle> domain = withColumnDomains(ImmutableMap.of(columnHandle, Domain.notNull(mapType)));
    MessageType fileSchema = new MessageType("hive_schema", new GroupType(OPTIONAL, "my_map", new GroupType(REPEATED, "map", new PrimitiveType(REQUIRED, INT32, "key"), new PrimitiveType(OPTIONAL, INT32, "value"))));
    Map<List<String>, RichColumnDescriptor> descriptorsByPath = getDescriptors(fileSchema, fileSchema);
    TupleDomain<ColumnDescriptor> tupleDomain = getParquetTupleDomain(descriptorsByPath, domain, fileSchema, true);
    assertTrue(tupleDomain.isAll());
}

Example 12 with RichColumnDescriptor

use of io.trino.parquet.RichColumnDescriptor in project trino by trinodb.

the class TestParquetPredicateUtils method testParquetTupleDomainStruct.

@Test
public void testParquetTupleDomainStruct() {
    RowType rowType = rowType(RowType.field("a", INTEGER), RowType.field("b", INTEGER));
    HiveColumnHandle columnHandle = createBaseColumn("my_struct", 0, HiveType.valueOf("struct<a:int,b:int>"), rowType, REGULAR, Optional.empty());
    TupleDomain<HiveColumnHandle> domain = withColumnDomains(ImmutableMap.of(columnHandle, Domain.notNull(rowType)));
    MessageType fileSchema = new MessageType("hive_schema", new GroupType(OPTIONAL, "my_struct", new PrimitiveType(OPTIONAL, INT32, "a"), new PrimitiveType(OPTIONAL, INT32, "b")));
    Map<List<String>, RichColumnDescriptor> descriptorsByPath = getDescriptors(fileSchema, fileSchema);
    TupleDomain<ColumnDescriptor> tupleDomain = getParquetTupleDomain(descriptorsByPath, domain, fileSchema, true);
    assertTrue(tupleDomain.isAll());
}

Example 13 with RichColumnDescriptor

use of io.trino.parquet.RichColumnDescriptor in project trino by trinodb.

the class TupleDomainParquetPredicate method matches.

@Override
public boolean matches(long numberOfRows, Map<ColumnDescriptor, Statistics<?>> statistics, ParquetDataSourceId id) throws ParquetCorruptionException {
    if (numberOfRows == 0) {
        return false;
    }
    if (effectivePredicate.isNone()) {
        return false;
    }
    Map<ColumnDescriptor, Domain> effectivePredicateDomains = effectivePredicate.getDomains().orElseThrow(() -> new IllegalStateException("Effective predicate other than none should have domains"));
    for (RichColumnDescriptor column : columns) {
        Domain effectivePredicateDomain = effectivePredicateDomains.get(column);
        if (effectivePredicateDomain == null) {
            continue;
        }
        Statistics<?> columnStatistics = statistics.get(column);
        if (columnStatistics == null || columnStatistics.isEmpty()) {
            // no stats for column
            continue;
        }
        Domain domain = getDomain(column, effectivePredicateDomain.getType(), numberOfRows, columnStatistics, id, timeZone);
        if (!effectivePredicateDomain.overlaps(domain)) {
            return false;
        }
    }
    return true;
}

Example 14 with RichColumnDescriptor

use of io.trino.parquet.RichColumnDescriptor in project trino by trinodb.

the class TupleDomainParquetPredicate method matches.

@Override
public boolean matches(long numberOfRows, ColumnIndexStore columnIndexStore, ParquetDataSourceId id) throws ParquetCorruptionException {
    requireNonNull(columnIndexStore, "columnIndexStore is null");
    if (numberOfRows == 0) {
        return false;
    }
    if (effectivePredicate.isNone()) {
        return false;
    }
    Map<ColumnDescriptor, Domain> effectivePredicateDomains = effectivePredicate.getDomains().orElseThrow(() -> new IllegalStateException("Effective predicate other than none should have domains"));
    for (RichColumnDescriptor column : columns) {
        Domain effectivePredicateDomain = effectivePredicateDomains.get(column);
        if (effectivePredicateDomain == null) {
            continue;
        }
        ColumnIndex columnIndex = columnIndexStore.getColumnIndex(ColumnPath.get(column.getPath()));
        if (columnIndex == null) {
            continue;
        }
        Domain domain = getDomain(effectivePredicateDomain.getType(), numberOfRows, columnIndex, id, column, timeZone);
        if (!effectivePredicateDomain.overlaps(domain)) {
            return false;
        }
    }
    return true;
}

Example 15 with RichColumnDescriptor

use of io.trino.parquet.RichColumnDescriptor in project trino by trinodb.

the class ParquetReader method initializeColumnReaders.

private void initializeColumnReaders() {
    for (PrimitiveColumnIO columnIO : columns) {
        RichColumnDescriptor column = new RichColumnDescriptor(columnIO.getColumnDescriptor(), columnIO.getType().asPrimitiveType());
        columnReaders[columnIO.getId()] = PrimitiveColumnReader.createReader(column, timeZone);
    }
}