Examples with StringStatisticsBuilder com.facebook.presto.orc.metadata.statistics.StringStatisticsBuilder used on opensource projects

Example 1 with StringStatisticsBuilder

use of com.facebook.presto.orc.metadata.statistics.StringStatisticsBuilder in project presto by prestodb.

the class ColumnWriters method createColumnWriter.

public static ColumnWriter createColumnWriter(int columnIndex, List<OrcType> orcTypes, Type type, ColumnWriterOptions columnWriterOptions, OrcEncoding orcEncoding, DateTimeZone hiveStorageTimeZone, DwrfEncryptionInfo dwrfEncryptors, MetadataWriter metadataWriter) {
    requireNonNull(type, "type is null");
    OrcType orcType = orcTypes.get(columnIndex);
    Optional<DwrfDataEncryptor> dwrfEncryptor = dwrfEncryptors.getEncryptorByNodeId(columnIndex);
    switch(orcType.getOrcTypeKind()) {
        case BOOLEAN:
            return new BooleanColumnWriter(columnIndex, type, columnWriterOptions, dwrfEncryptor, metadataWriter);
        case FLOAT:
            return new FloatColumnWriter(columnIndex, type, columnWriterOptions, dwrfEncryptor, metadataWriter);
        case DOUBLE:
            return new DoubleColumnWriter(columnIndex, type, columnWriterOptions, dwrfEncryptor, metadataWriter);
        case BYTE:
            return new ByteColumnWriter(columnIndex, type, columnWriterOptions, dwrfEncryptor, metadataWriter);
        case DATE:
            checkArgument(orcEncoding != DWRF, "DWRF does not support %s type", type);
            return new LongColumnWriter(columnIndex, type, columnWriterOptions, dwrfEncryptor, orcEncoding, DateStatisticsBuilder::new, metadataWriter);
        case SHORT:
            return new LongColumnWriter(columnIndex, type, columnWriterOptions, dwrfEncryptor, orcEncoding, IntegerStatisticsBuilder::new, metadataWriter);
        case INT:
        case LONG:
            if (columnWriterOptions.isIntegerDictionaryEncodingEnabled() && orcEncoding == DWRF) {
                // ORC V1 does not support Integer Dictionary encoding. DWRF supports Integer dictionary encoding.
                return new LongDictionaryColumnWriter(columnIndex, type, columnWriterOptions, dwrfEncryptor, orcEncoding, metadataWriter);
            }
            return new LongColumnWriter(columnIndex, type, columnWriterOptions, dwrfEncryptor, orcEncoding, IntegerStatisticsBuilder::new, metadataWriter);
        case DECIMAL:
            checkArgument(orcEncoding != DWRF, "DWRF does not support %s type", type);
            return new DecimalColumnWriter(columnIndex, type, columnWriterOptions, orcEncoding, metadataWriter);
        case TIMESTAMP:
            return new TimestampColumnWriter(columnIndex, type, columnWriterOptions, dwrfEncryptor, orcEncoding, hiveStorageTimeZone, metadataWriter);
        case BINARY:
            return new SliceDirectColumnWriter(columnIndex, type, columnWriterOptions, dwrfEncryptor, orcEncoding, BinaryStatisticsBuilder::new, metadataWriter);
        case CHAR:
            checkArgument(orcEncoding != DWRF, "DWRF does not support %s type", type);
        // fall through
        case VARCHAR:
        case STRING:
            if (columnWriterOptions.isStringDictionaryEncodingEnabled()) {
                return new SliceDictionaryColumnWriter(columnIndex, type, columnWriterOptions, dwrfEncryptor, orcEncoding, metadataWriter);
            }
            int stringStatisticsLimit = columnWriterOptions.getStringStatisticsLimit();
            return new SliceDirectColumnWriter(columnIndex, type, columnWriterOptions, dwrfEncryptor, orcEncoding, () -> new StringStatisticsBuilder(stringStatisticsLimit), metadataWriter);
        case LIST:
            {
                int fieldColumnIndex = orcType.getFieldTypeIndex(0);
                Type fieldType = type.getTypeParameters().get(0);
                ColumnWriter elementWriter = createColumnWriter(fieldColumnIndex, orcTypes, fieldType, columnWriterOptions, orcEncoding, hiveStorageTimeZone, dwrfEncryptors, metadataWriter);
                return new ListColumnWriter(columnIndex, columnWriterOptions, dwrfEncryptor, orcEncoding, elementWriter, metadataWriter);
            }
        case MAP:
            {
                ColumnWriter keyWriter = createColumnWriter(orcType.getFieldTypeIndex(0), orcTypes, type.getTypeParameters().get(0), columnWriterOptions, orcEncoding, hiveStorageTimeZone, dwrfEncryptors, metadataWriter);
                ColumnWriter valueWriter = createColumnWriter(orcType.getFieldTypeIndex(1), orcTypes, type.getTypeParameters().get(1), columnWriterOptions, orcEncoding, hiveStorageTimeZone, dwrfEncryptors, metadataWriter);
                return new MapColumnWriter(columnIndex, columnWriterOptions, dwrfEncryptor, orcEncoding, keyWriter, valueWriter, metadataWriter);
            }
        case STRUCT:
            {
                ImmutableList.Builder<ColumnWriter> fieldWriters = ImmutableList.builder();
                for (int fieldId = 0; fieldId < orcType.getFieldCount(); fieldId++) {
                    int fieldColumnIndex = orcType.getFieldTypeIndex(fieldId);
                    Type fieldType = type.getTypeParameters().get(fieldId);
                    fieldWriters.add(createColumnWriter(fieldColumnIndex, orcTypes, fieldType, columnWriterOptions, orcEncoding, hiveStorageTimeZone, dwrfEncryptors, metadataWriter));
                }
                return new StructColumnWriter(columnIndex, columnWriterOptions, dwrfEncryptor, fieldWriters.build(), metadataWriter);
            }
    }
    throw new IllegalArgumentException("Unsupported type: " + type);
}