Examples with ParquetDictionary com.facebook.presto.hive.parquet.dictionary.ParquetDictionary used on opensource projects

Example 1 with ParquetDictionary

use of com.facebook.presto.hive.parquet.dictionary.ParquetDictionary in project presto by prestodb.

the class TupleDomainParquetPredicate method getDomain.

@VisibleForTesting
public static Domain getDomain(Type type, ParquetDictionaryDescriptor dictionaryDescriptor) {
    if (dictionaryDescriptor == null) {
        return null;
    }
    ColumnDescriptor columnDescriptor = dictionaryDescriptor.getColumnDescriptor();
    Optional<ParquetDictionaryPage> dictionaryPage = dictionaryDescriptor.getDictionaryPage();
    if (!dictionaryPage.isPresent()) {
        return null;
    }
    ParquetDictionary dictionary;
    try {
        dictionary = dictionaryPage.get().getEncoding().initDictionary(columnDescriptor, dictionaryPage.get());
    } catch (Exception e) {
        // OK to ignore exception when reading dictionaries
        return null;
    }
    int dictionarySize = dictionaryPage.get().getDictionarySize();
    if (type.equals(BIGINT) && columnDescriptor.getType() == PrimitiveTypeName.INT64) {
        List<Domain> domains = new ArrayList<>();
        for (int i = 0; i < dictionarySize; i++) {
            domains.add(Domain.singleValue(type, dictionary.decodeToLong(i)));
        }
        domains.add(Domain.onlyNull(type));
        return Domain.union(domains);
    } else if (type.equals(BIGINT) && columnDescriptor.getType() == PrimitiveTypeName.INT32) {
        List<Domain> domains = new ArrayList<>();
        for (int i = 0; i < dictionarySize; i++) {
            domains.add(Domain.singleValue(type, (long) dictionary.decodeToInt(i)));
        }
        domains.add(Domain.onlyNull(type));
        return Domain.union(domains);
    } else if (type.equals(DOUBLE) && columnDescriptor.getType() == PrimitiveTypeName.DOUBLE) {
        List<Domain> domains = new ArrayList<>();
        for (int i = 0; i < dictionarySize; i++) {
            domains.add(Domain.singleValue(type, dictionary.decodeToDouble(i)));
        }
        domains.add(Domain.onlyNull(type));
        return Domain.union(domains);
    } else if (type.equals(DOUBLE) && columnDescriptor.getType() == PrimitiveTypeName.FLOAT) {
        List<Domain> domains = new ArrayList<>();
        for (int i = 0; i < dictionarySize; i++) {
            domains.add(Domain.singleValue(type, (double) dictionary.decodeToFloat(i)));
        }
        domains.add(Domain.onlyNull(type));
        return Domain.union(domains);
    } else if (isVarcharType(type) && columnDescriptor.getType() == PrimitiveTypeName.BINARY) {
        List<Domain> domains = new ArrayList<>();
        for (int i = 0; i < dictionarySize; i++) {
            domains.add(Domain.singleValue(type, Slices.wrappedBuffer(dictionary.decodeToBinary(i).getBytes())));
        }
        domains.add(Domain.onlyNull(type));
        return Domain.union(domains);
    }
    return null;
}