protected ColumnReader(
ParquetRecordReader parentReader,
int allocateSize,
ColumnDescriptor descriptor,
ColumnChunkMetaData columnChunkMetaData,
boolean fixedLength,
V v,
SchemaElement schemaElement)
throws ExecutionSetupException {
this.parentReader = parentReader;
this.columnDescriptor = descriptor;
this.columnChunkMetaData = columnChunkMetaData;
this.isFixedLength = fixedLength;
this.schemaElement = schemaElement;
this.valueVec = v;
this.pageReader =
new PageReader(
this, parentReader.getFileSystem(), parentReader.getHadoopPath(), columnChunkMetaData);
if (columnDescriptor.getType() != PrimitiveType.PrimitiveTypeName.BINARY) {
if (columnDescriptor.getType() == PrimitiveTypeName.FIXED_LEN_BYTE_ARRAY) {
dataTypeLengthInBits = columnDescriptor.getTypeLength() * 8;
} else {
dataTypeLengthInBits = ParquetRecordReader.getTypeLengthInBits(columnDescriptor.getType());
}
}
}
protected ColumnReader(
ParquetRecordReader parentReader,
int allocateSize,
ColumnDescriptor descriptor,
ColumnChunkMetaData columnChunkMetaData,
boolean fixedLength,
ValueVector v)
throws ExecutionSetupException {
this.parentReader = parentReader;
this.columnDescriptor = descriptor;
this.columnChunkMetaData = columnChunkMetaData;
this.isFixedLength = fixedLength;
if (allocateSize > 1) {
valueVecHolder = new VectorHolder(allocateSize, v);
} else {
valueVecHolder = new VectorHolder(5000, v);
}
this.pageReadStatus =
new PageReadStatus(
this, parentReader.fileSystem, parentReader.hadoopPath, columnChunkMetaData);
if (columnDescriptor.getType() != PrimitiveType.PrimitiveTypeName.BINARY) {
dataTypeLengthInBits = ParquetRecordReader.getTypeLengthInBits(columnDescriptor.getType());
}
}
/**
* start a column inside a block
*
* @param descriptor the column descriptor
* @param valueCount the value count in this column
* @param statistics the statistics in this column
* @param compressionCodecName
* @throws IOException
*/
public void startColumn(
ColumnDescriptor descriptor, long valueCount, CompressionCodecName compressionCodecName)
throws IOException {
state = state.startColumn();
if (DEBUG) LOG.debug(out.getPos() + ": start column: " + descriptor + " count=" + valueCount);
currentEncodings = new HashSet<parquet.column.Encoding>();
currentChunkPath = ColumnPath.get(descriptor.getPath());
currentChunkType = descriptor.getType();
currentChunkCodec = compressionCodecName;
currentChunkValueCount = valueCount;
currentChunkFirstDataPage = out.getPos();
compressedLength = 0;
uncompressedLength = 0;
// need to know what type of stats to initialize to
// better way to do this?
currentStatistics = Statistics.getStatsBasedOnType(currentChunkType);
}